Content uploaded by Sarah Fiona Barker
Author content
All content in this area was uploaded by Sarah Fiona Barker on Jan 16, 2018
Content may be subject to copyright.
A probabilistic quantitative microbial risk
assessment model of norovirus disease burden
from wastewater irrigation of vegetables in
Shepparton, Australia
Hoi-Fei Mok
a
, S. Fiona Barker
b
, Andrew J. Hamilton
a,
*
a
Department of Agriculture and Food Systems, Melbourne School of Land and Environment, University of Melbourne,
Parkville, VIC 3010, Australia
b
Department of Resource Management and Geography, Melbourne School of Land and Environment, University of
Melbourne, Parkville, VIC 3010, Australia
article info
Article history:
Received 4 November 2013
Received in revised form
24 January 2014
Accepted 26 January 2014
Available online 6 February 2014
Keywords:
Agriculture
Diarrhea
Horticulture
Public health
QMRA
abstract
Wastewater can be an important resource for water-scarce regions of the world, but a
major barrier to its use is the associated health risk. Quantitative microbial risk assessment
(QMRA) is a probabilistic modeling technique used to determine the health risks from
wastewater reuse, but only a handful of QMRA studies have examined the norovirus health
risks from consumption of vegetables irrigated with human wastewater, even though
norovirus is a, if not the most, significant microbial cause of diarrheal disease world-wide.
Furthermore, the majority of these studies have focused only on risks from lettuce con-
sumption. To meet the knowledge gap in health risks for other vegetables, a QMRA model
was constructed for agricultural wastewater irrigation in the regional city of Shepparton,
Australia, using fecal shedding rates to estimate norovirus concentration in raw sewage.
Annual norovirus disease burden was estimated for the consumption of lettuce, broccoli,
cabbage, Asian vegetables, and cucumber after irrigation with treated wastewater. Results
indicate that the waste stabilization pond treatment did not have sufficient virus removal
to meet the World Health Organization (WHO) threshold for acceptable level of risk for
wastewater reuse, but addition of disinfection treatments provided acceptable results for
consumption of cucumber and broccoli. This is the first QMRA study to incorporate virus
accumulation from previous wastewater irrigation events.
ª2014 Elsevier Ltd. All rights reserved.
1. Introduction
Increasing rainfall variability and extreme weather events,
coupled with growing human population needs, has amplified
pressure on water resources. Reuse of treated urban waste-
water can play a significant role in meeting water demands
and increasing water security, with schemes ranging from
agricultural uses such as crop and pasture irrigation to toilet
and laundry reuse (Hamilton et al., 2007). However, concern
*Corresponding author. Tel.: þ3 8344 9308.
E-mail address: andrewjh@unimelb.edu.au (A.J. Hamilton).
Available online at www.sciencedirect.com
ScienceDirect
journal homepage: www.elsevier.com/locate/watres
water research 54 (2014) 347e362
0043-1354/$ esee front matter ª2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.watres.2014.01.060
about the microbial quality of wastewater limits its wide-
spread use (Hamilton et al., 2007; Toze, 2006).
Enteric viruses are particularly concerning due to their
high shedding rate and high persistence in the environment
(Carter, 2005), and have been responsible for gastroenteritis
outbreaks associated with consumption of contaminated
water (Greer et al., 2009; Kukkula et al., 1999; ter Waarbeek
et al., 2010). Norovirus is a leading cause of non-bacterial
gastroenteritis affecting all age groups world-wide, aided by
its resistance to treatment, ability to survive in water, and
fecal-oral transmission route (Lodder and Husman, 2005;
Matthews et al., 2012; Widdowson et al., 2005). With their
nonenveloped structure, similar to other human enteric vi-
ruses like poliovirus and echovirus, noroviruses show resis-
tance to environmental degradation such as pH changes and
desiccation (Cannon et al., 2006; D’Souza et al., 2006). Evalu-
ations of healthcare-associated gastroenteritis outbreaks
found that norovirus was detected in 74% of the outbreaks in
Edinburgh, Scotland from 2007 to 2009 and 63% of outbreaks in
England from 2002 to 2003, costing the respective National
Health Services V1.2 million and 184 million USD (Danial et al.,
2011; Lopman et al., 2004). The large number of norovirus-
related illnesses per year can result in substantial costs to
society and therefore it is an important pathogen to consider.
One approach to estimating microbiological risks, as rec-
ommended by the World Health Organization (WHO) guide-
lines and Australian Guidelines for Water Recycling, for
wastewater reuse incorporates the quantitative microbial risk
assessment (QMRA) technique with Monte Carlo simulation
(NRMMC, 2006; WHO, 2006). In brief, QMRA translates the
pathogen dose that the consumer is exposed to for a particular
scenario into probabilities of infection and illness through
four steps: hazard identification, exposure assessment,
human doseeresponse effects, and final risk characterization
(Haas et al., 1999). Such risk assessments are useful for
informing regulatory bodies governing wastewater reuse
whether or not reuse schemes are safe for the populations
exposed (NRMMC, 2006).
Although QMRA can be a powerful tool in risk estimation,
construction of the model typically requires a multitude of
data regarding pathogen density, rate of pathogen-ingestion
exposure, and pathogen doseeresponse relationship. There
has only been a small number of studies that assessed
norovirus risk since the development of the norovirus
doseeresponse model (Teunis et al., 2008), and of these only
a few that focused on risk from treated sewage (Mara and
Sleigh, 2010a; Schoen et al., 2011; Soller et al., 2010).
Furthermore, vegetables modeled for wastewater irrigation
risks have largely been limited to lettuce (Barker et al.,
2013b; Petterson et al., 2001; Seidu et al., 2008; Shuval
et al., 1997). Considering this knowledge gap for different
vegetable risks and the growing prevalence of wastewater
reuse world-wide, particularly for agricultural use (Hamilton
et al., 2007), it is important to investigate the potential nor-
ovirus disease burden from wastewater irrigation of
different vegetables, including Asian vegetables such as
choy sum and bok choy, in order to develop suitable risk
management plans.
This study focuses on the potential risks from wastewater
agricultural irrigation in the regional Australian city of
Shepparton, Victoria, located 160 km north of Melbourne.
Shepparton is located in the heart of the Goulburn Valley,
known as Australia’s food bowl with its multitude of food
processing centers, dairy farmers, and other agribusinesses.
The major food processing company located in Shepparton,
SPC Ardmona, accounts for more than 80% of the average
daily wastewater inflow to the Shepparton wastewater
treatment plant (WTP) (GVW, 2004). Wastewater reuse has
been widely implemented across the capital cities around the
country since the recent decade-long drought and interest is
growing to expand reuse in the inland regional cities
(Radcliffe, 2010). With the anticipated population growth,
Shepparton is projected to have an increase of over 5000 ML/
year in water usage by 2060 (GVW, 2012). However, the recent
Basin Plan regarding water bodies in the Murray-Darling
Basin, including the major Goulburn River, has outlined re-
ductions in future river water diversions for agriculture
(Murray-Darling Basin Authority, 2012), thus creating the
need to investigate alternative water sources. Goulburn Val-
ley Water, the wastewater management company overseeing
the Shepparton WTP, is particularly interested in exploring
new wastewater reuse schemes for agriculture and horti-
culture, which is in line with local farmers’ desire to diversify
their production beyond fruit to vegetables (RMCG and GSCC,
2012).
Unlike for many reuse schemes in major cities of Australia,
financial resources are a significant constraint for wastewater
reuse in regional cities (Australian Centre for Water Recycling
Excellence, 2013). The Shepparton WTP utilizes wastewater
stabilization ponds (WSP), a low-cost technology, for primary
and secondary treatment. The Shepparton WTP also uses the
Actiflo process, a method of chemically enhanced high-rate
sedimentation, for tertiary treatment of wastewater to be
discharged into the Goulburn River during times of high flow,
but this is a relatively expensive treatment method compared
to other technologies and is designed more for removal of
suspended solids and organic matter rather than pathogens
(Plum et al., 1998; Scherrenberg, 2006). Since cost is a factor, it
is important to determine whether the WSP treatment alone
can satisfy health targets for reuse.
Given the knowledge gaps for norovirus and different
vegetable risks associated with wastewater irrigation, the
need for alternative water sources in Shepparton, and the
local interest in pursuing wastewater reuse for agriculture,
this study aims to determine the norovirus disease burden
from wastewater agricultural irrigation in Shepparton for
lettuce, broccoli, cabbage, cucumber, bok choy, choy sum, and
gai lan. One key challenge in this study is the paucity of data,
particularly for norovirus concentrations. With the method-
ological difficulties in norovirus detection and high costs
associated with current measurement techniques (Atmar,
2010), there are few published datasets for norovirus in
treated or untreated wastewater and no available data for
norovirus concentrations in Australian treatment plants. In
lieu of experimental data, this model utilized the human fecal
shedding method to estimate norovirus sewage concentra-
tions (Barker et al., 2013b). An additional aim of this study was
to investigate whether low-cost technology such as WSP can
meet guideline health targets for wastewater reuse or whether
further treatment is needed.
water research 54 (2014) 347e362348
Table 1 eModel input parameters including general distributions and fit parameters.
Model parameter Units Distribution type (values)
a
w[Mean]
b
References and fit statistics
F¼fecal weight from people
with infectious diarrhea
g feces/person/day Lognormal 3 (7.09, 0.10, 1003.7) e
truncated at zero w[218.8]
Bytzer et al. (1990)
S
N
¼norovirus shedding
rate in feces
No. viruses/g, assuming
that genomic copies
equal number of viruses
10
PERT
(9.2,11,12.2) w[1.65 101
1
]Atmar et al. (2008)
O
G
¼daily per capita incidence
of gastroenteritis in Australia
Probability Mixture w[2.52 10
3
], see Table 3 Hall et al. (2004)
O
N
¼proportion of norovirus-related
cases per gastroenteritis case in
Australia
Proportion 0.3948 Hall et al. (2005)
P¼total population in Shepparton Person 29,553 ABS (2012)
W
T
¼daily total wastewater volume
at the Shepparton WTP
L/day Mixture w[1.79 10
7
], see Table 3 GVW (2013)
R
WSP
¼removal of viruses through
waste stabilization pond treatment
log
10
unit Uniform (0, 4) w[1.99] Da Silva et al. (2008);
El-Deeb Ghazy et al.
(2008); Oragui et al.
(1995); Oragui et al. (1987)
R
AT
¼removal of viruses through
advanced treatment
log
10
unit
Actiflo Uniform (0, 3) w[1.50] Sigmund et al. (2006)
Chlorination Uniform (0.75, 4) w[2.38] Francy et al. (2012); Rose
et al. (1996); Sobsey (1989);
Tree et al. (2003, 2005)
Ozone Uniform (0.5, 4) w[2.25] Harakeh and Butler
(1984); Joret et al. (1982);
Sobsey
(1989); Xu et al. (2002)
Ultraviolet (UV) irradiation Uniform (0.25, 4) w[2.11] Francy et al. (2012);
Oppenheimer et al.
(1997); Tree et al.
(2003, 2005)
B¼body mass of people in
Shepparton
kg-person Mixture w[63.61] ABS (1998a,b, 2012);
DoHA (2008)
k¼in-field virus kinetic decay
constant
days
1
Normal (1.07, 0.07) etruncated
at zero w[1.07]
Petterson et al. (2001, 2002)
t¼withholding period Days Uniform (0, 2) w[1.00]
W
P
¼proportion of population
that wash vegetables prior to
consumption
Proportion Mixture w[0.87] Barker et al. (2013a,b);
Mitakakis et al. (2004)
R
W
¼reduction of virus
concentration by post-harvest
vegetable washing
log
10
unit PERT (0.1, 1, 2) w[1.02] Bae et al. (2011); Baert et al.
(2008, 2009); Barker et al.
(2013a,b); Croci et al. (2002);
Dawson et al. (2005);
Fraisse et al. (2011)
Norovirus doseeresponse and
illness parameters
Hypergeometric Beta Poisson
a¼0.040,
b¼0.055, a¼0.9997, n¼0.00255,
r¼0.086
Teunis et al. (2008)
D¼norovirus disease burden DALY/case of illness Uniform (3.71 10
4
, 6.23 10
3
)
w[3.30 10
3
]
Barker et al. (2013a,b);
Cressey and Lake (2009);
Haagsma et al. (2008);
Havelaar et al. (2012);
Kemmeren et al. (2006);
Lake et al. (2010);
Verhoef et al. (2013)
S
F
¼susceptibility fraction Proportion Uniform (0.8, 1.0) w[0.9] Atmar (2010); Barker et al.
(2013a,b); Le Pendu et al.
(2006); Rydell et al. (2011);
Thorven et al. (2005)
a
Distribution types and values: Log normal3 (m,s, location), where population parameters ^mand b
sare calculated as follows: ^m¼lnx0:5lnð1þ
ðs=xÞ2Þand b
s¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
lnð1þðs=xÞ2Þ
q;where xis the sample mean and sis the sample standard deviation (Limpert et al., 2001), and location is a shift
parameter; Mixture ecombination of various distributions; Normal (mean, sd); PERT (minimum, mode, maximum); and Uniform (minimum,
maximum).
b
Mean values calculated after 10,000 iterations.
water research 54 (2014) 347e362 349
2. Methods
2.1. Exposure model
The scenarios modeled in this study involve the consump-
tion of vegetables irrigated with wastewater treated at the
Shepparton WTP, specifically raw consumption of bok choy
(Brassica rapa var. chinensis), gai lan (Brassica oleracea var.
alboglabra), choy sum (Brassica rapa var. parachinensis), lettuce
(Lactuca sativa), broccoli (Brassica oleracea var. italica), cu-
cumber (Cucumis sativus), and cabbage (Brassica oleracea var.
capitata). These vegetables were chosen because there is in-
terest in diversifying horticultural production beyond stone
fruit and pomefruit to vegetables, such as Asian vegetables,
which are becoming more popular in Australia (RIRDC, 2011;
RMCG and GSCC, 2012). Norovirus was chosen as the mi-
crobial hazard to model because it is the most common
cause of gastroenteritis in Melbourne, the capital city of
Victoria, Australia (Sinclair et al., 2005). The norovirus dose
the consumer is exposed to on the kth iteration of the 365
days of the year (l
k
; viruses ingested/person/day) was
defined as
lk¼X7
x¼0
CWTP
1000 VekðtþxÞMB;(1)
where C
WTP
is the concentration of norovirus in the treated
wastewater (virus/L), Vis the volume of irrigation water
caught by the crop (mL/g), Mis the population average daily
consumption of vegetable across all days of the year per capita
per kg of body mass (g/kg-person/day), Bis the human body
mass (kg-person), kis the in-field virus kinetic decay constant
(day
1
), xis the number of days of past irrigation events, and t
is the withholding time between the last wastewater irrigation
event and harvest (days). General fit parameters for all prob-
ability distributions are shown in Table 1 and vegetable-
specific fit parameters are shown in Table 2. This paper
considered only overhead sprinkler irrigation because that is
the typical irrigation method for vegetables such as lettuce in
the region.
Norovirus and other enteric viruses are known to be
resistant to environmental degradation (Cannon et al., 2006;
D’Souza et al., 2006), so it is important to take into account
viral accumulation. This model assumes daily irrigation
events, which is the common commercial practice for leafy
vegetables, and also assumes the worst-case scenario that all
wastewater from irrigation evaporates without washing away
accumulated viruses. One week’s worth of past irrigation
events (x¼7) were included along with one current irrigation
event (x¼0). Using the maximum sewage concentration, a
deterministic evaluation of viral decay using Eq. (1) showed a
7 log
10
reduction after 7 days and this was deemed sufficient
to account for viral accumulation on plant surfaces.
The water capture factors (V) for lettuce, gai lan, bok choy,
and choy sum were taken from Mok and Hamilton (2014). The
original data for the broccoli and cabbage volume distribu-
tions were taken from Hamilton et al. (2006) and probability
density functions were fitted through @Risk software (Palisade
Corporation, Newfield, New York). Using the goodness-of-fit
statistics from @Risk, the probability density functions that
Table 2 eModel input parameters relating to vegetables, including distributions and fit parameters.
Model parameter Units Distribution type (values)
a
w[Mean]
b
References and fit statistics
V¼volume of water
captured by vegetable
mL/g
Bok choy Lognormal (3.79, 0.37) w[2.41 10
2
]Mok and Hamilton (2014)
Broccoli Lognormal (4.07, 0.40) w[1.84 10
2
]Hamilton et al. (2006);
Chi Sq ¼11.54 AIC ¼707.70
c
Cabbage (Savoy King
or Grand Slam)
Lognormal 3 (3.71, 0.72, 0.004) w[3.49 10
2
]Hamilton et al. (2006);
Chi Sq ¼2.24 AIC ¼144.81
c
Choy sum Lognormal (3.08, 0.48) w[5.10 10
2
](Mok and Hamilton (2014)
Cucumber Normal (0.0036, 0.0012) etruncated
at zero w[3.61 10
3
]
Hamilton et al. (2006); Shuval
et al. (1997)
Gai lan Lognormal 3 (2.99, 0.45, 0.005) w[6.08 10
2
]Mok and Hamilton (2014)
Lettuce (Green oak) Lognormal 3 (4.57, 0.50, 0.006) w[1.78 10
2
]Mok and Hamilton (2014)
M¼per capita consumption
of vegetable
g/kg-person/day
Bok choy Lognormal (3.13, 1.67) w[0.06] US EPA (2003)
Broccoli Lognormal (5.27, 2.37) w[0.08] AusVeg (2010)
Cabbage Lognormal (3.94, 2.08) w[0.19] ABS (1998a); US EPA (2011)
Choy sum Lognormal (3.13, 1.67) w[0.06] US EPA (2003)
Cucumber Lognormal (5.98, 2.52) w[0.07] ABS (1998a); US EPA (2011)
Gai lan Lognormal (3.13, 1.67) w[0.06] US EPA (2003)
Lettuce Lognormal (2.23, 1.62) w[0.42] AusVeg (2011)
a
Distribution types and values: log normal (m,s), where population parameters ^mand b
sare calculated as follows: ^m¼lnx0:5lnð1þðs=xÞ2Þ;and
b
s¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
lnð1þðs=xÞ2Þ
q;, where xis the sample mean and sis the sample standard deviation (Limpert et al., 2001), Log normal 3 (m,s, locatio n), where
location is a shift parameter; Normal (mean, sd).
b
Mean values calculated after 10,000 iterations.
c
AIC ¼Akaike information criterion.
water research 54 (2014) 347e362350
gave the lowest fit statistics and best graphical fit were
selected for the distributions. The distribution for volume
capture on cucumber was also taken from Hamilton et al.
(2006), but it should be noted that the original data for cu-
cumber water capture were obtained via submersion experi-
ments (Shuval et al., 1997) and likely represent the worst-case
irrigation scenario. A conservative approach was taken for all
vegetables wherein it was assumed that all pathogens in the
wastewater captured on the vegetable attach to its surface
(Hamilton et al., 2006; Petterson et al., 2001; Shuval et al., 1997).
Consumption rates were assumed to be lognormally
distributed, as assumed by others (Barker, 2013; Ma et al.,
2003). Per capita Australian consumption data for lettuce,
broccoli, cabbage, and cucumber were obtained from AusVeg
(AusVeg, 2010, 2011) and the Australian Bureau of Statistics
(ABS, 1998a), but no Australian estimates of standard devi-
ation were available. It was assumed that standard deviation
in consumption behavior in the USA would be similar to that
in Australia and therefore, the measures of standard devia-
tion for the US data for lettuce, broccoli, cabbage, and cu-
cumber were used. All these vegetables had the same
standard deviation, 1.44 g/kg-person/day, as derived from
the 1998 US EPA CSFII survey results (US EPA, 2011). To
match the units of standard deviation, the mean Australian
consumption rates (g/person/day) were divided by mean
body mass in Shepparton (63.6 kg-person; weighted by age
and sex). No Australian consumption data were accessible
for Asian vegetables, thus the US consumption data for dark
green vegetables from the US EPA CSFII survey was used. To
reduce overestimation, it was assumed that dark green
vegetable consumption was composed only of bok choy,
choy sum, and gai lan consumption and thus the individual
vegetable consumption was calculated by dividing the values
from the dark green vegetable lognormal distribution by 3.
Since the consumption rate of Asian vegetables in Australia
is unknown, this portion of the assessment can be seen as
screening level.
The Mixture distribution for body mass in Shepparton was
constructed from several different data sources and body
mass distribution was assumed to be lognormal (Ni Mhurchu
et al., 2004; Walls et al., 2010). Using Australian body mass
statistics (arithmetic mean and standard deviation) for chil-
dren (DoHA, 2008) and adults (ABS, 1998b), distributions were
constructed for each age range and sex, weighted by propor-
tion of the population in Shepparton (ABS, 2012). Random
samples (n¼100,000) of weighted body mass for each age and
sex range were selected for an overall Mixture distribution of
the population weighted mean body mass. This distribution of
Bwas randomly drawn from with replacement for the calcu-
lation of l
k
.
The in-field virus decay was assumed to be Normally
distributed, following experiments on Bacteroides fragilis
bacteriophage B40-8 on lettuce (Petterson et al., 2001, 2002).
Post-harvest decay of virus was considered negligible (Badawy
et al., 1985) and therefore was not included in the model.
Post-harvest vegetable washing practices have been shown
to reduce viral loads on vegetables (Baert et al., 2008, 2009).
The norovirus dose that the consumer is exposed to after
reduction of viral load from washing (l
wash
; virus ingested/
person/day) was defined as
lwash ¼X7
x¼0
CWTP
1000 VekðtþxÞMB 10Rw(2)
where R
W
is the log
10
reduction in virus concentration from
vegetable washing. In previous studies, viral reduction from
washing produce with tap water was reported as between 0.1
and 2 log
10
units (Bae et al., 2011; Baert et al., 2008, 2009; Croci
et al., 2002; Dawson et al., 2005; Fraisse et al., 2011), with five
studies reporting a mean reduction of 1 0.2 log
10
unit; this
was used as the most likely value in the PERT distribution for
R
W
(Table 1).
2.2. Estimation of norovirus concentration
The estimation of norovirus concentration in wastewater is
difficult, complicated by a number of factors, including poor
viral detection efficiency, expensive test methods, and no
available cell culture to determine infectivity (Atmar, 2010).
Use of indicator organisms to estimate norovirus concentra-
tions would be problematic due to a multitude of factors
affecting this correlation, such as time of year, carriage rates,
environmental stressors, growth of organisms, and transport
characteristics (Ahmad et al., 2009; Feachem et al., 1983; Wu
et al., 2011). To avoid this complication, this study used
human fecal shedding rates to estimate norovirus concen-
tration in wastewater.
The norovirus load in raw sewage (C
Sewage
; virus/L) was
estimated as
CSewage ¼FSNOGONP
WT
;(3)
where Fis the weight of feces from people with diarrhea (g
feces/person/day), S
N
is the norovirus shedding rate (no. vi-
ruses/g feces), O
G
is the daily per capita incidence of gastro-
enteritis in Australia, O
N
is the proportion of norovirus-related
gastroenteritis cases in Australia, Pis the population of
Shepparton (persons), and W
T
is the daily total wastewater
volume at the Shepparton WTP (L/day). It was assumed that
diarrhea would be the predominant pathway for pathogens to
enter wastewater. The three-parameter truncated lognormal
distribution of Fwas constructed from the fecal weights of
patients with infectious diarrhea (Bytzer et al., 1990). The PERT
distribution for norovirus shedding rate in feces (S
N
) was
constructed based on results of an experimental human
infection trial for norovirus (Atmar et al., 2008).
Given the seasonal variation in the volume of the SPC
cannery wastewater output and thus the volume of raw
sewage inflow at the Shepparton WTP, the daily wastewater
volume (W
T
; L/day) was calculated separately for each month
of the year as
WT¼Monthly raw sewage inflow
days per month ;(4)
where the average monthly raw sewage inflow (L/month;
GVW, 2013) and days per month are reported in Table 3. Each
month of the year was represented proportionately in the
model simulations by sampling each monthly module only
once for every yearly estimation of C
Sewage
.
Hall et al. (2004) determined the population-level incidence
of gastroenteritis in a national Australian survey of 6087
water research 54 (2014) 347e362 351
households across all seven states and territories. The survey
participant responses and resulting gastroenteritis incidences
were recorded in 4-week intervals and weighted by age, sex,
household size, and geographic location using census data.
The resulting seasonal gastroenteritis incidence rates (cases
of gastroenteritis/person/season) were used in this QMRA
model, where the distribution of values was assumed to be
Normal (Table 3). Daily incidence of gastroenteritis (O
G
; cases
of gastroenteritis/person/day) was estimated as
OG¼Seasonal gastroenteritis incidence
days per season ;(5)
where days per season was defined as the number of days per
three-month season (Table 3). Since the model simulation was
broken down into monthly modules, a given season-specific
O
G
was expressed for three individual monthly modules
before being replaced by the next season-specific O
G
to reflect
the variation in gastroenteritis incidence. The proportion of
norovirus-related gastroenteritis cases (O
N
) is a point estimate
based on a study of gastroenteritis caused by known patho-
gens in Australia (Hall et al., 2004, 2005).
The norovirus load in the treated wastewater (C
WTP
; virus/
L) is given by two equations:
CWTP ¼CSewage10RWSP (6)
CWTP ¼CSewage10RWSP 10RAT (7)
where R
WSP
is the removal of viruses by WSPs (log
10
unit), and
R
AT
is the removal of viruses by advanced treatment (log
10
unit). Eq. (6) represents scenarios where WSPs only are used to
treat the wastewater. Since there is a possibility that WSP
treatment is insufficient to satisfy the health target of 10
6
disability-adjusted life years (DALY)/person/year for tolerable
risk from wastewater reuse, Eq. (7) represents scenarios where
advanced treatments such as Actiflo, chlorination, ozonation,
or ultraviolet (UV) irradiation are used as additional treatment
to WSPs.
There are few data available on norovirus removal by
WSPs. Virus survival in aquatic environments such as WSPs
depends on a number of factors, including temperature, solar
radiation, pH, adsorption, and predation by bacteria and pro-
tozoa (Allwood et al., 2003; Feachem et al., 1983; Feng et al.,
2003; Frederick and Lloyd, 1995; Maynard et al., 1999; Shapiro
et al., 2010; Sinton et al., 2002; Watts et al., 1995). Therefore a
Uniform distribution was constructed based on data compiled
from several international studies of virus removal by WSPs
(Da Silva et al., 2008; da Silva et al., 2007; El-Deeb Ghazy et al.,
2008; Locas et al., 2010; Oragui et al., 1995, 1987).
Although the Actiflo plant is the tertiary treatment
technology available at Shepparton WTP, it was not
designed for virus removal. However, there is evidence that
it can remove up to 3 log
10
of coliphage (Sigmund et al.,
2006). In the absence of other data, this was used for the
Uniform distribution for Actiflo treatment. Three common
disinfection treatments were also considered in this model:
chlorination, ozonation, and UV irradiation. The removal
efficiency depends on the wastewater quality, the sensitivity
of the target microorganisms to treatment, and various
other conditions (Hijnen et al., 2006; Lazarova et al., 1999;
Sobsey, 1989; Tree et al., 2003). To simulate this variation,
generalized Uniform distributions of virus log removal from
these treatments were constructed based on results of pre-
vious studies (Francy et al., 2012; Harakeh and Butler, 1984;
Joret et al., 1982; Oppenheimer et al., 1997; Rose et al.,
1996; Tree et al., 2003, 2005; Xu et al., 2002). For all treat-
ments, the lower end of the Uniform distributions repre-
sents scenarios with poor treatment conditions or treatment
failure while the upper end represents those with ideal
treatment conditions.
2.3. Doseeresponse model and risk characterization
The norovirus doseeresponse models constructed by Teunis
et al. (2008) were used along with the fit parameters for the
combined inocula dataset (8fIIa þ8fIIb). The combined
dataset allows for any degree of virus aggregation and in the
absence of information on aggregation state, it is therefore the
best doseeresponse model in this situation. Modeling the
Table 3 eSeasonality of raw sewage inflow at Shepparton WTP and gastroenteritis incidence in Australia.
Month Days per
month
Mean monthly raw sewage inflow at
Shepparton WTP from 1991 to 2013
(L/month)
a
Gastroenteritis incidence across
Australia per season (case/person/season)
b
December 31 5.3298 10
8
Summer eNormal (0.27, 0.0255)
c
January 31 5.7409 10
8
February 28 6.7189 10
8
March 31 7.4688 10
8
Autumn eNormal (0.18, 0.0306)
April 30 5.5729 10
8
May 31 5.0418 10
8
June 30 4.4829 10
8
Winter eNormal (0.25, 0.0459)
July 31 5.1225 10
8
August 31 5.1212 10
8
September 30 4.6309 10
8
Spring eNormal (0.22, 0.0306)
October 31 5.0843 10
8
November 30 4.9002 10
8
a
GVW (2013).
b
Hall et al. (2004), weighted by state, age, sex, and household size.
c
Normal (mean, sd).
water research 54 (2014) 347e362352
probability of infection for this dataset normally uses the
Gauss hypergeometric function,
2
F
1
¼(a,b;c;z), from Eq. (4) of
Teunis et al. (2008), which has the constraint jzj<1(Wolfram
Research, 2013). Since the fit parameters of Teunis et al. give
jzj¼3,332, the Pfaff transformation (Weisstein, 2013) was
applied to Teunis et al.’s model to give
pðinfjdoseÞ¼12
42F1b;lwashð1aÞ
a;aþb;a 1
1alwashð1aÞ
a3
5;
(8)
where aand bare fit parameters, and ais the fit parameter of
the logarithmic series aggregate size distribution. As noted in
Barker et al. (2013b), the Pfaff transformation fails for doses
greater than 33,323 viruses ingested/person/day and the full
Beta Poisson model provides an adequate approximation
(Teunis, pers. comm., 19 January 2012). Thus,
when l>33;323;pðinfjdoseÞ¼11F1ða;aþb;lÞ(9)
where
1
F
1
is the Kummer confluent hypergeometric function
(Teunis et al., 2008; Teunis, pers. comm., 19 January 2012).
The conditional probability of illness after infection (p
(illjinf)
;
person
1
day
1
) was determined following Teunis et al.’s
(2008) model as
pðilljinfÞ¼1ð1þnlwashÞr(10)
where nand rare scale and shape parameters respectively for
the Gamma distribution of the duration of infection, as
determined by Teunis et al. (2008). The probability of illness
per dose (p
(illjdose)
; person
1
day
1
) was then calculated as
pðill jdoseÞ¼pðinfjdoseÞpðilljdoseÞ:(11)
2.3.1. Population-level risk
The dose calculations (Eq. (1) and Eq. (2)) were developed for
the individuals consuming wastewater-irrigated vegetables
without cooking (consumer-only risk). The risk to the average
person in Shepparton (population-level risk; p
popn
; person-
1
day
1
) was estimated as
ppopn ¼p1WpþpwashWp;(12)
where pis the daily probability of illness per dose (p
(illjdose)
) for
individuals who do not wash vegetables prior to consumption,
p
wash
is the probability for individuals who wash vegetables
prior to consumption, and W
P
is the proportion of the popu-
lation that washes vegetables prior to consumption.
The proportion of the population that washes vegetables
prior to consumption was calculated using results from a
survey of over 500 households in Melbourne on vegetable
washing behavior (Mitakakis et al., 2004). The survey asked
participants whether they washed salads and/or vegetables
before serving. An estimated washing probability was
assigned to each of the qualitative survey choices as described
in Barker et al. (2013b). Briefly, the estimated washing proba-
bilities ranged from 0 to 1 and were weighted by the propor-
tion of population giving that qualitative survey answer. The
sum of the different washing probabilities gave W
P
. It was
assumed that the population washing behaviors in Melbourne
would be similar to the population washing behaviors in
Shepparton.
2.3.2. Annual risk
The daily probabilities of illness were used as inputs for the
annual probability of illness calculation, P, which is given as
P¼1Y365
k¼11pk(13)
where p
k
is the probability of illness for the kth iteration of the
365 days in a year, and where events are assumed to be in-
dependent (Karavarsamis and Hamilton, 2010). Lettuce is
grown year-round and many of the other modeled vegetables
are grown the majority of the year, so it was assumed that
wastewater irrigation of vegetables and consumption of those
vegetables occurs year-round. The average daily consumption
rate for vegetables (M) accounts for the variability in vegetable
consumption, including non-consumption days, across the
year. This means parameterizing the number of consumption
events is unnecessary for this model, as all 365 days of the
year must be modeled to reflect possible consumption.
Annual disease burden was calculated using the DALY
metric, expressed as the number of years lost due to illness,
disability or premature death. The annual disease burden (A;
DALY/person/year) for norovirus illness was estimated as
A¼PðilljdoseÞDSF;(14)
where P
(illjdose)
is the annual probability of illness per dose
(from Eq. (13)), Dis the norovirus disease burden (DALY/case
of norovirus illness), and S
F
is the proportion of the population
susceptible to the disease. Since there is no norovirus disease
burden estimate for Australia, international estimates were
used to construct a Uniform distribution for D(Haagsma et al.,
2008; Havelaar et al., 2012; Kemmeren et al., 2006; Lake et al.,
2010; Verhoef et al., 2013). Although there is evidence of ge-
netic resistance to norovirus infection from histo-blood group
antigens and secretor status (Lindesmith et al., 2003; Teunis
et al., 2008), the variation between norovirus genotypes sug-
gests that it is possible that every person would be genetically
susceptible to at least one norovirus genotype (Atmar 2010).
The doseeresponse models constructed by Teunis et al. (2008)
were based on secretor-positive volunteers, in other words,
individuals susceptible to norovirus infection. Theoretically
the model should apply only to this portion of the population,
which accounts for up to 80% of white populations (Le Pendu
et al., 2006; Rydell et al., 2011; Thorven et al., 2005). But since
susceptibility to norovirus is uncertain, this study used
methods cited in Barker et al. (2013b) and represented S
F
as a
Uniform distribution from secretor-positive individuals (0.8)
to all individuals (1.0).
2.4. Model implementation
To account for uncertainty and variability in the parameters,
Monte Carlo simulation of 3,650,000 iterations was used to
calculate daily probabilities, thus providing enough values for
10,000 annual probability estimates. Each iteration involved
drawing a set of values from the input parameter probability
distributions (Tables 1 and 2). For every simulation of annual
probability, daily probabilities were selected proportionally
water research 54 (2014) 347e362 353
from the appropriate months of the year such that a distri-
bution of annual risk values was constructed. All modeling
and analysis were implemented through R version 3.0.1 (The R
Foundation for Statistical Computing, 2013). For all model
scenarios, 90% confidence intervals were calculated using the
percentile method (Buckland, 1984).
Spearman rank order correlation was conducted after
10,000 iterations of the exposure model, comparing the un-
certainty relationships between the input variables and p
popn
.
This correlation technique was preferred to linear correlation
because most of the relationships were non-linear.
3. Results
Overall, consumption of wastewater-irrigated cucumber and
broccoli posed the least risks while consumption of lettuce
posed the greatest risks. In scenarios with WSP-only waste-
water treatment, the median annual norovirus disease burden
across the different vegetables ranged from 7.95 10
5
to
2.34 10
3
DALY/person/year. The 90% confidence intervals
for the DALYs for the vegetables with WSP treatment are as
follows: bok choy [1.62 10
4
, 2.84 10
3
], broccoli
[7.57 10
5
, 2.43 10
3
], cabbage [3.38 10
4
, 3.88 10
3
],
choy sum [3.07 10
4
, 2.84 10
3
], cucumber [2.37 10
6
,
7.04 10
4
], gai lan [3.39 10
4
, 3.81 10
3
], and lettuce
[4.66 10
4
, 4.40 10
3
]. The disinfection and Actiflo treat-
ment scenarios produced very similar annual disease bur-
dens, with 1 or 2 orders of magnitude difference between the
Actiflo, chlorine, ozone, and UV treatments. The median
annual disease burdens from the WSP þdisinfection sce-
narios were on average 1e2 orders of magnitude smaller than
those from the WSP-only scenarios, with values ranging from
5.95 10
8
to 8.63 10
4
DALY/person/year. The mean virus
concentration in raw sewage estimated by the model was
6.03 10
7
virus/L. The median probability of infection for
washers ranged from 2.94 10
8
to 1.51 10
3
while the
median probability of illness per dose ranged from 1.35 10
16
to 3.52 10
7
. For non-washers, the probabilities were one
order of magnitude higher, with the median probability of
infection ranging from 3.0810
7
to 1.52 10
2
and median
probability of illness per dose ranging from 2.22 10
15
to
5.34 10
6
. The population-level risk of illness per dose
ranged from 2.60 10
15
to 6.13 10
6
across the different
scenarios.
The box-and-whiskers plots of annual disease burden es-
timates show that the WSP-only scenarios spanned a range
from 10
5
to 10
3
DALY/person/year, with all of the median
values exceeding the WHO (2006) and Australian (NRMMC,
2006) guidelines threshold of 10
6
DALY/person/year for
acceptable level of risk from wastewater reuse (Fig. 1). How-
ever, of the advanced treatment scenarios, most of the cu-
cumber consumption scenarios have median disease burden
values that meet the guideline threshold (Fig. 1). Several of the
broccoli and bok choy scenarios have median disease burden
values slightly above the 10
6
DALY/person/year threshold.
Overall, lettuce consumption posed the greatest risks while
cucumber consumption posed the least risks.
The sensitivity analyses revealed that uncertainty in the
virus removal by WSP, R
WSP
, was the most influential factor of
all the variables affecting variability in the population-level
risk for daily probability of illness per dose, p
popn
, with
Spearman rank order correlation coefficients ranging from
0.651 to 0.731 (Table 4). Uncertainty in the other wastewater
treatment parameters, shedding rate, and consumption rate,
also produced moderately strong correlations. Uncertainty in
the kinetic decay constant, k, did not significantly contribute
to variation and was the only parameter with coefficients not
significantly different from zero.
Fig. 1 eBox-and-whiskers graphs of annual norovirus
disease burden from consumption of wastewater-irrigated
vegetables. Boxes represent the inter-quartile range, the
solid lines within the boxes are medians, the whiskers are
the 10th and 90th percentiles, the dots are the 5th and 95th
percentiles, and the small dashed lines are the arithmetic
means. The large dashed lines denote the guideline
threshold for acceptable annual disease burden.
water research 54 (2014) 347e362354
4. Discussion
In water-scarce areas such as Australia, wastewater can be an
important resource. However, before wastewater reuse
schemes are implemented, issues such as health risks must
be considered. This study is the first to investigate viral dis-
ease burden from proposed wastewater reuse for a regional
Australian city and the first QMRA to incorporate virus accu-
mulation from prior irrigation events. Of the vegetables
modeled, cucumber posed the least risks while lettuce posed
the greatest risks. While none of the WSP-only scenarios
satisfied the WHO and Australian guidelines in terms of
acceptable levels of risk from wastewater reuse, consumption
of cucumber, with the addition of chlorine, ozone, or UV, met
the 10
6
DALY/person/year threshold. A number of the other
scenarios were slightly above the threshold. Not only does this
have important implications for Shepparton stakeholders
interested in wastewater agricultural irrigation, but also for
farmers, water resource managers, and policymakers
throughout the world.
The borderline cases of broccoli and bok choy under the
chlorine treatment as well as broccoli with ozone treatment
deserve further investigation. The median annual disease
burdens associated with these scenarios were within the
10
6
DALY/person/year order of magnitude, but by conserva-
tive measures, the median values should be less than the
threshold to be considered acceptable. Enhancing the path-
ogen reduction through non-treatment means such as longer
withholding periods or replacing the last irrigation event with
potable water may allow the estimated disease burden to drop
below the threshold. Further study can elucidate if these non-
treatment risk management techniques can have an impact
on the resulting disease burdens.
In this study all the WSP-only scenarios failed to meet the
threshold for acceptable risk levels, indicating that the virus
removal capabilities of this low-cost technology alone were
insufficient. Although non-treatment management tech-
niques can be applied in conjunction with WSP, the magni-
tude of the annual disease burdens from the WSP-only
scenarios indicate that a considerable risk reduction of 3 or 4
orders of magnitude is needed. In countries with lower nor-
ovirus incidence rates, use of WSP alone may be sufficient, but
with the high incidence rates in Australia, use of WSP and
further advanced treatment is needed for safe wastewater
irrigation. Further study would be needed to see if non-
treatment techniques could achieve this reduction, but it is
possible that the required non-treatment options may not be
practicable. For example, the length of the withholding period
might exceed plant water requirements or the number of
replacement irrigation events might be so numerous, in
comparison to the wastewater irrigation days that it would be
logistically not worth even using wastewater. In addition,
non-treatment methods may not meaningfully reduce risk in
comparison to treatment methods. Indeed, the 1-log reduc-
tion from washing vegetables prior to consumption is modest
in comparison to the 3 to 4-log reduction by WSP and other
treatment technologies. While arguably the combination of
several non-treatment methods may equate to reduction by
improved treatment techniques, the difficulty in changing
behavior around washing vegetables or withholding irrigation
needs to be taken into consideration.
In making the decision of which advanced treatment sys-
tem to use, several different factors must be considered.
Although the Actiflo infrastructure is already in place at the
Shepparton WTP, its pathogen removal ability was insuffi-
cient to meet the guideline threshold. The disinfection tech-
niques all had similar pathogen removal abilities, though
chlorination had marginally better removal rates, so other
factors should be considered in determining the most suitable
treatment technique. Ozonation is a high-cost process with
operational difficulties, creating hazardous by-products, and
often requires high amounts of ozone for effective pathogen
inactivation (Gehr et al., 2003; Lazarova et al., 1999). This op-
tion offers no advantage over the others in this case and is not
preferred by the wastewater management company. While
UV irradiation has cost-effective disinfection efficiency for
bacteria and viruses without using or producing toxic chem-
icals, it is only effective within the right turbidity range
(Hijnen et al., 2006). In cases of low turbidity, UV irradiation
would be a useful disinfection process, but it is unlikely that
turbidity conditions would be met each time, thus further
treatment would be needed to ensure suitable wastewater
turbidity before disinfection. The other option is to use chlo-
rination, a low-cost disinfection technique and a well under-
stood technology that effectively inactivates bacteria and
viruses (WHO, 2006). However, chlorination creates toxic by-
products from the use of hazardous chemicals, which may
have detrimental impacts on receiving water ecosystems (Szal
et al., 1991), although this can be addressed through dechlo-
rination after disinfection (Lazarova et al., 1999). Economic
analyses would be needed to compare the costs of UV
irradiation-turbidity removal treatment versus chlor-
inationedechlorination. Empirical studies would also be
required to determine the requisite levels of chlorine and UV
irradiation for norovirus disinfection in Shepparton
wastewater.
Table 4 eSensitivity analyses for probability of
illness per dose at the population-level (p
popn
).
Values represent Spearman rank order
correlation coefficients (r) for input variables in
relation to p
popn
and are ranked based on
strength of correlation.
Parameters
a
Coefficients
R
WSP
0.651e0.731
R
AT UV
0.507e0.568
M0.418e0.581
R
AT Ozone
0.476e0.531
R
AT Chlorine
0.453e0.501
R
AT Actiflo
0.418e0.480
SR
N
0.287e0.335
R
W
0.193 to 0.220
F0.159e0.186
t0.128 to 0.170
B0.100e0.124
V0.076e0.116
O
G
0.042e0.076
W
T
0.022e0.048
k0.013 to 0.040
a
Input parameters defined in Tables 1 and 2.
water research 54 (2014) 347e362 355
One important factor to consider is system reliability.
Variability in pathogen removal ability of the treatment sys-
tem has a substantial influence on the resulting estimates of
annual disease burden. Using a point estimate of 4 log
10
removal by a hypothetical treatment system, which is less
than the combined mean log reduction values from WSP
combined with any of the advanced treatments, we evaluated
the impact on annual disease burden from lettuce consump-
tion, which had posed the highest risk amongst all the sce-
narios. The result was a median annual disease burden of
6.47 10
7
DALY/person/year, which meets the health target.
This suggests that, given appropriate operating conditions,
the modeled treatment technology is capable of sufficient
viral removal for acceptable levels of risk. If the WSP tech-
nology were operating in ideal conditions such that the
maximum removal of 4 log
10
(Table 1) was reached consis-
tently, the disinfection treatment would not be needed.
However, with the variation in treatment effectiveness with
poor wastewater quality and the possibility of system break-
down, which is accounted for in this model by the lower end
of the treatment parameter distributions, there is a lack of
consistency in the removal capabilities and over the course of
10,000 model simulations, the estimated annual disease bur-
dens exceed the guideline threshold. If treatment effective-
ness can be streamlined through correlating WSP hydraulic
conditions and geometry to virus removal or reducing
turbidity for enhanced disinfection effectiveness (Persson and
Wittgren, 2003; Saqqar and Pescod, 1995), this may improve
system reliability and increase the likelihood of the estimated
disease burden meeting the guideline threshold.
While the inclusion of virus accumulation from previous
irrigation events is novel for QMRA studies, the model pre-
sented here is a simplified worst-case scenario. Pathogen
attachment to and survival on vegetable surfaces depends on a
number of pathogen surface properties (e.g. isoelectric point
and hydrophobicity), environmental parameters (e.g. pH, ionic
strength, and temperature), and produce characteristics (e.g.
surface structure and charge) (Brandl, 2006; Deboosere et al.,
2012; Kukavica-Ibrulj et al., 2004; Langlet et al., 2008). This
may result in differential attachment across crops, as found
with Norwalk virus-like particles localizing to veins of romaine
lettuce, but binding non-specifically to iceberg lettuce (Gandhi
et al., 2010). Attachment can also be dose dependent (Gandhi
et al., 2010) and time dependent (Deboosere et al., 2012). It is
also possible that attached particles can subsequently detach
from the crop surface and/or be washed off by irrigation water
or rainfall. Viral decay also differs between pathogens types
(Stine et al., 2005). In the absence of decay data for norovirus,
the kinetic decay constant for this model came from Bacteroides
fragilis bacteriophage B40-8, which is more resistant relative to
other pathogens to decay in environmental conditions and
therefore is a conservative model for human enteric viruses
(Lucena et al., 1996; Petterson and Ashbolt, 2001). By assuming
no virus detachment across all crops and using a conservative
viral decay model, the viral accumulation in this model may be
overestimated. Further modeling incorporating the above
mentioned parameters would be needed to get a more com-
plete picture.
Another important consideration is the uncertainty in the
doseeresponse and illness models. While Teunis et al. (2008)
offer the only doseeresponse relationship constructed from
known norovirus inocula, there are considerable weaknesses,
including a small dataset and a marginal fit of the
doseeresponse relationship. A recent study gives another set
of norovirus doseeresponse relationships from contaminated
oyster consumption (Thebault et al., 2013). Thebault et al.’s
(2013) study utilized an indirect method of calculating dose
by using a Poisson-gamma mixture distribution to represent
the number of noroviruses in oysters. We believe the Teunis
et al. (2008) doseeresponse relationship is more appropriate
for our study because it involves known inocula and offers a
model for no assumed aggregate state, while the Poisson-
gamma mixture distribution assumes homogenous and fully
dispersed inocula. Many other norovirus studies have also
utilized the Teunis et al. (2008) doseeresponse relationship
(Barker, 2013; Barker et al., 2013a,b; Mara and Sleigh, 2010b;
Schoen et al., 2011; Soller et al., 2010). However, to explore
whether this doseeresponse model underestimates risk, we
re-ran the QMRA model for the WSP scenario using fit pa-
rameters for Seþ/GI and Seþ/GII from Thebault et al. (2013).
For both Seþ/GI and Seþ/GII, the median probability of
infection and illness per dose was similar, ranging from
2.90 10
3
to 4.80 10
1
to 1.39 10
5
to 2.27 10
1
,
respectively, for washers and 2.91 10
2
to 6.92 10
1
and
2.02 10
4
to 6.30 10
2
, respectively, for non-washers,
several orders of magnitude greater than those found in
using the Teunis et al. (2008) parameters. Due to the
maximum value of 1 for the annual probability of illness,
these higher daily probabilities resulted in effectively the
same DALY for all scenarios, a median of 2.97 10
3
DALY/
person/year. This doseeresponse model predicts that con-
sumption of wastewater-irrigated vegetables in the Sheppar-
ton context would result in at least one infection per person
per year, thus implying that use of Teunis et al.’s (2008)
doseeresponse relationship markedly underestimates the
risks of infection and illness. Although the doseeresponse
model from Teunis et al. (2008) is likely the best choice
because it is based on a challenge study, the Thebault et al.
(2013) study demonstrates the uncertainty around these
values and highlights the possibility of underestimation.
This study made no assumption for virus aggregate form and
used the doseeresponse equation from Teunis et al. (2008) cor-
responding to the combination of the aggregated “8flla” and non-
aggregated “8fllb” datasets. However, the aggregation parameter
in Teunis et al.’s (2008) model (a¼0.9997) was still retained for
use in the equation, despite inclusion of non-aggregate data in
the model. This non-aggregate “8fllb” data came from a fresh
norovirus inocula while the aggregate “8flla” data came from
inoculastored away for more than 25 years. While fresh sewage
may contain non-aggregate virus forms, it is possible that
wastewater treatment may cause virus aggregation. Since the
assumption of aggregate form can have a considerable effect on
the results (Ashbolt et al., 2010; Soller et al., 2010), further
microbiological research is needed to elucidate this.
The sensitivity analyses showed that uncertainty in the
treatment processes had the most influence on the probability
of infection (Table 4), likely because these parameters were
among the least site-specific. The actual rate of norovirus
removal by WSP or disinfection depends heavily on the
wastewater quality and other conditions (Hijnen et al., 2006;
water research 54 (2014) 347e362356
Lazarova et al., 1999; Sobsey, 1989; Tree et al., 2003), so without
empirical data from Shepparton WTP, the uncertainty in these
parameters will be large. Consumption rate was also not ideal
with a lack of information on Asian vegetable consumption in
Australia. With no general estimate of leafy/dark green vege-
table consumption in Australia, the consumption rate of dark
green vegetables in the USA was used as a proxy for the
combined consumption of Asian vegetables. To gage whether
dividing the dark green vegetable consumption by 3 was an
underestimation for the individual consumption rate, several
of the Asian vegetable scenarios were run with the full dark
green vegetable consumption distribution. The resulting
annual disease burdens were around the same order of
magnitude, some 1 order of magnitude larger. This indicates
that the model is only mildly sensitive to potential variation
between the full and divided dark green vegetable consump-
tion rate. Although this does not indicate if the USA dark green
vegetable consumption rate is an accurate reflection of
Australian Asian vegetable consumption, the assumption that
the distribution represents the combined consumption has
negligible effect on the output. In constructing biologically
relevant QMRAs, the approach used to set up the exposure
model is critical. Using E. coli as an indicator organism for vi-
ruses, as historically done in many QMRA studies, is prob-
lematic, as it makes the following assumptions: that all
sources of fecal contamination are human, that there is non-
differential inactivation of E. coli and viruses in the treatment
system, and that there is a linear relationship between E. coli
and norovirus concentrations. These problems are avoided
with the fecal loading approach as applied in this QMRA
model. By modeling the norovirus concentrations from the
point of raw sewage using norovirus shedding rates, this study
specifically accounted for the human norovirus inputs and
seasonality of norovirus infection at a community level. This
method of estimating fecal load has been successfully used in
other QMRA studies (Barker et al., 2013a,b; Ottoson and
Stenstrom, 2003). Use of pathogen shedding rates for estima-
tion of pathogen concentrations can enable QMRA modeling
in situations where direct pathogen sampling data are un-
available; this is particularly useful for viruses considering the
high costs associated with test methods and poor detection
efficiency (Atmar, 2010). The disadvantages of such an
approach are the high number of input data necessary for
constructing the exposure model, increased complexity in
modeling the entire pathway from raw sewage to irrigation
water, and potentially greater uncertainty in the final estimate
of pathogen concentrations.
Similar results were found when comparing the estimated
norovirus concentration in sewage from the shedding rate
method in this study to published municipal sewage data.
This model gave a mean of 6.03 10
7
viruses/L in raw sewage
using the shedding rate method, which is within one order of
magnitude of previously reported sewage concentrations in a
number of different countries (Aw and Gin, 2010; Haramoto
et al., 2006; Katayama et al., 2008; La Rosa et al., 2010;
Laverick et al., 2004; Nordgren et al., 2009; van den Berg
et al., 2005). However, the argument that the shedding rate
method may be overestimating the pathogen concentrations
in raw sewage should still be seriously considered. The
shedding rate method utilizes several worst case or
conservative assumptions, such as no viral decay between
household discharge and wastewater treatment plant and
100% virus viability and infectivity. Both assumptions may
result in an overestimation of norovirus concentrations in
sewage. Additionally, there are large uncertainties associated
with each model parameter, both epistemic and aleatory, and
they can be difficult to quantify.
With the absence of empirical sewage concentration data,
as in the model presented in this study, the alternative method
is to rely on published municipal sewage data from other
countries. But this involves another series of assumptions. The
norovirus-related gastroenteritis incidence rates in different
countries vary drastically, from 0.013 cases/person/year in En-
gland (Flint et al., 2005) to 0.082e0.113 cases/person/year in
Australia (Hall et al., 2004,2006; Sinclair et al., 2005). Evidence of
similar incidence rates and other conditions such as disease
seasonality and population size would be needed to justify
using data from another country. Virus recovery rates from
molecular methods vary drastically. Studies comparing
different molecular extraction methods for viruses from large
water bodies or sewage have found 1e3 orders of magnitude
difference amongst the methods (Belguith et al., 2006; Fumian
et al., 2010; Rutjes et al., 2005). The recovery efficiency would
need to be accounted for in the model if such data are used.
Finally, inputs to sewers may differ, especially in regard to the
ratio of industrial to municipal effluent, which would influence
the wastewater quality. Without consideration of these, a
model based on literature values could seriously under-
estimate the sewage concentration by 1e3 orders of magnitude
(Barker, 2013; Silverman et al., 2013). Overriding these problems
is the fact that there are only two studies of norovirus con-
centration in sewage that report recovery efficiency and both
are from Japan (Haramoto et al., 2006; Katayama et al., 2008).
Barker (2013) focused exclusively on the issue of estimating
sewage concentrations for Melbourne, Australia through
different methods for QMRA. The study compared results from
the shedding rate method to published municipal sewage
concentrations from Japan. It found that the shedding rate
method produced norovirus concentrations that were
approximately3 orders of magnitude higher than the published
Japanese datasets and 2 orders of magnitude higher than the
modified dataset Barker (2013). This shows that the shedding
rate method is the more conservative estimate of sewage
concentrations than the published literature. Given the lack of
information on the prevalence of norovirus-related gastroen-
teritis in Japan, it is inappropriate to assume the same sewage
concentrations would exist in Japan and in Shepparton to
justify use of literature values. With these findings, the shed-
ding rate method may be the best estimate of local conditions.
5. Conclusions
To assess the health risks associated with wastewater reuse in
Shepparton, a QMRA model was used to estimate the annual
norovirus disease burden from consumption of vegetables
irrigated with treated wastewater. Major findings were:
(1) None of the WSP-only scenarios met the threshold of
10
6
DALY/person/year for acceptable level of risk.
water research 54 (2014) 347e362 357
This indicates that for the Australian context, the nor-
ovirus removal capabilities of WSPs are insufficient and
further advanced treatment would be necessary.
(2) Consumption of cucumber with WSP and any of the
advanced treatments met the 10
6
DALY/person/year
threshold. Several other scenarios were borderline,
indicating that WSP with disinfection treatment pro-
vided sufficient pathogen reduction for some vegetable
types. The cost-benefit of chlorine and UV irradiation
treatments should be investigated next to determine
the most suitable treatment for the Shepparton WTP.
(3) A point estimate of 4 log
10
removal representing the
mean combined treatment virus removal gave an
acceptable level of risk for lettuce consumption, indi-
cating that the WSP and disinfection treatment tech-
nologies have the capability of meeting the guideline
threshold even with the highest risk vegetable scenario.
However, streamlining of system reliability and treat-
ment efficiency would be needed to consistently reach
this 4 log
10
removal.
(4) Comparison of estimates of the norovirus concentra-
tion in raw sewage from this study with published
municipal sewage data revealed that the concentrations
were within 1 order of magnitude, indicating that the
fecal shedding method provided a realistic estimate.
(5) Current washing practices appear to make a relatively
modest contribution to reducing risk.
(6) Although the Teunis et al. (2008) doseeresponse rela-
tionship is likely the best choice for modeling because of
known norovirus inocula and equations for no assumed
aggregate state, compared to the indirect method for
calculating the relationship by Thebault et al. (2013), the
marginal fit from the small dataset is a considerable
weakness. Comparison of model results that incorpo-
rated the norovirus doseeresponse model constructed
by Thebault et al. (2013) suggests that the Teunis et al.
(2008) doseeresponse relationship may be under-
estimating infection and illness risks.
(7) A more complex model for virus attachmente
detachmenteaccumulation is needed.
Acknowledgments
We would like to acknowledge the advice given by Joanne
O’Toole on microbiological matters and the helpful comments
made by three anonymous reviewers, which greatly
strengthened the paper.
references
ABS, 1998a. 4315.0 eApparent Consumption of Selected
Foodstuffs, Australia, Preliminary, 1997e98. Australian Bureau
of Statistics, Canberra, Australia.
ABS, 1998b. 4359.0 eHow Australians Measure up, 1995.
Australian Bureau of Statistics, Canberra, Australia.
ABS, 2012. 3235.0 ePopulation by Age and Sex, Regions of
Australia. Australian Bureau of Statistics, Canberra, Australia.
Ahmad, F., Tourlousse, D.M., Stedtfeld, R.D., Seyrig, G.,
Herzog, A.B., Bhaduri, P., Hashsham, S.A., 2009. Detection and
occurrence of indicator organisms and pathogens. Water
Environ. Res. 81 (10), 959e980.
Allwood, P.B., Malik, Y.S., Hedberg, C.W., Goyal, S.M., 2003.
Survival of F-specific RNA coliphage, feline calicivirus, and
Escherichia coli in water: a comparative study. Appl. Environ.
Microbiol. 69 (9), 5707e5710.
Ashbolt, N.J., Schoen, M.E., Soller, J.A., Rose, D.J., 2010. Predicting
pathogen risks to aid beach management: the real value of
quantitative microbial risk assessment (QMRA). Water Res. 44
(16), 4692e4703.
Atmar, R.L., 2010. Noroviruses: state of the art. Food Environ.
Virol. 2 (3), 117e126.
Atmar, R.L., Opekun, A.R., Gilger, M.A., Estes, M.K., Crawford, S.E.,
Neill, F.H., Graham, D.Y., 2008. Norwalk virus shedding after
experimental human infection. Emerg. Infect. Dis. 14 (10),
1553e1557.
Australian Centre for Water Recycling Excellence, 2013. Removing
the Barriers for Water Recycling in Regional and Remote
Australia eWorkshop Summary. Asia Pacific Recycled Water
Conference. Australian Centre for Water Recycling Excellence,
Brisbane, Australia.
AusVeg, 2010. Vegetable Spotlight: Broccoli. AusVeg, Australia.
AusVeg, 2011. Vegetable Spotlight: Lettuce. AusVeg, Australia.
Aw, T.G., Gin, K.Y.H., 2010. Environmental surveillance and
molecular characterization of human enteric viruses in tropical
urban wastewaters. J. Appl. Microbiol. 109 (2), 716e730.
Badawy, A.S., Gerba, C.P., Kelley, L.M., 1985. Survival of rotavirus
SA-11 on vegetables. Food Microbiol. 2 (3), 199e205.
Bae, J.Y., Lee, J.S., Shin, M.H., Lee, S.H., Hwang, I.G., 2011. Effect of
wash treatments on reducing human norovirus on iceberg
lettuce and Perilla leaf. J. Food Prot. 74 (11), 1908e1911.
Baert, L., Uyttendaele, M., Vermeersch, M., Van Coillie, E.,
Debeverei, J., 2008. Survival and transfer of murine norovirus
1, a surrogate for human noroviruses, during the production
process of deep-frozen onions and spinach. J. Food Prot. 71 (8),
1590e1597.
Baert, L., Vandekinderen, I., Devlieghere, F., Van Coillie, E.,
Debevere, J., Uyttendaele, M., 2009. Efficacy of sodium
hypochlorite and peroxyacetic acid to reduce murine
norovirus 1, B40-8, Listeria monocytogenes, and Escherichia coli
O157:H7 on shredded iceberg lettuce and in residual wash
water. J. Food Prot. 72 (5), 1047e1054.
Barker, S.F., 2013. Risk of norovirus gastroenteritis from
consumption of vegetables irrigated with highly treated
municipal wastewater eevaluation of methods to estimate
sewage quality. Risk. Anal. http://dx.doi.org/10.1111/risa.12138.
Barker, F., Packer, M., Scales, P., Gray, S., Snape, I., Hamilton, A.J.,
2013a. Pathogen reduction requirements for direct potable
reuse in Antarctica: evaluating human health risks in small
communities. Sci. Total Environ. 461e462, 723e733.
Barker, S., O’Toole, J., Sinclair, M., Malawaraarachchi, M.,
Leder, K., Hamilton, A., 2013b. A probabilistic model in
norovirus disease burden associated with greywater irrigation
of lettuce in households in Melbourne, Australia. Water Res.
47 (3), 1421e1432.
Belguith, K., Hassen, A., Aouni, M., 2006. Comparative study of
four extraction methods for enterovirus recovery from
wastewater and sewage sludge. Bioresour. Technol. 97 (3),
414e419.
Brandl, M.T., 2006. Annu. Rev. Phytopathol., 367e392.
Buckland, S.T., 1984. Monte-Carlo confidence intervals.
Biometrics 40 (3), 811e817.
Bytzer, P., Stokholm, M., Andersen, I., Lundhansen, B.,
Demuckadell, O.B.S., 1990. Etiology, medical history, and fecal
weight in adult patients referred for diarrhea ea prospective
survey. Scand. J. Gastroenterol. 25 (6), 572e578.
water research 54 (2014) 347e362358
Cannon, J.L., Papafragkou, E., Park, G.W., Osborne, J., Jaykus, L.-A.,
Vinje, J., 2006. Surrogates for the study of norovirus stability
and inactivation in the environment: a comparison of murine
norovirus and feline calicivirus. J. Food Prot. 69 (11),
2761e2765.
Carter, M.J., 2005. Enterically infecting viruses: pathogenicity,
transmission and significance for food and waterborne
infection. J. Appl. Microbiol. 98 (6), 1354e1380.
Cressey, P., Lake, R., 2009. Risk Ranking: DALY Estimates for
Selected Foodborn Diseases in New Zealand Using Revised
Dutch Disability Weights. Institute of Environmental Science
& Research Limited, Christchurch Science Centre, New
Zealand.
Croci, L., De Medici, D., Scalfaro, C., Fiore, A., Toti, L., 2002. The
survival of hepatitis A virus in fresh produce. Int. J. Food
Microbiol. 73 (1), 29e34.
D’Souza, D.H., Sair, A., Williams, K., Papafragkou, E., Jean, J.,
Moore, C., Jaykus, L., 2006. Persistence of caliciviruses on
environmental surfaces and their transfer to food. Int. J. Food
Microbiol. 108 (1), 84e91.
Da Silva, A.K., Le Guyader, F.S., Le Saux, J.C., Pommepuy, M.,
Montgomery, M.A., Elimelech, M., 2008. Norovirus removal
and particle association in a waste stabilization pond. Environ.
Sci. Technol. 42 (24), 9151e9157.
da Silva, A.K., Le Saux, J.C., Parnaudeau, S., Pommepuy, M.,
Elimelech, M., Le Guyader, F.S., 2007. Evaluation of removal of
noroviruses during wastewater treatment, using real-time
reverse transcription-PCR: different behaviors of genogroups I
and II. Appl. Environ. Microbiol. 73 (24), 7891e7897.
Danial, J., Cepeda, J.A., Cameron, F., Cloy, K., Wishart, D.,
Templeton, K.E., 2011. Epidemiology and costs associated with
norovirus outbreaks in NHS Lothian, Scotland 2007e2009. J.
Hosp. Infect. 79 (4), 354e358.
Dawson, D.J., Paish, A., Staffell, L.M., Seymour, I.J., Appleton, H.,
2005. Survival of viruses on fresh produce, using MS2 as a
surrogate for norovirus. J. Appl. Microbiol. 98 (1), 203e209.
Deboosere, N., Pinon, A., Caudrelier, Y., Delobel, A., Merle, G.,
Perelle, S., Temmam, S., Loutreul, J., Morin, T., Estienney, M.,
Belliot, G., Pothier, P., Gantzer, C., Vialette, M., 2012. Adhesion
of human pathogenic enteric viruses and surrogate viruses to
inert and vegetal food surfaces. Food Microbiol. 32 (1), 48e56.
DoHA, 2008. The 2007 National Children’s Nutrition and Physical
Activity Survey. Department of Health and Ageing,
Commonwealth of Australia.
El-Deeb Ghazy, M.M., El-Senousy, W.M., Abdel-Aatty, A.M.,
Kamel, M., 2008. Performance evaluation of a waste
stabilization pond in a rural area in Egypt. Am. J. Environ. Sci.
4 (4), 316e326.
Feachem, R.G., Bradley, D.J., Garelick, H., Mara, D.D., 1983.
Sanitation and Disease: Health Aspects of Excreta and
Wastewater Management. John Wiley & Sons, Chichester, UK,
pp. 3e22.
Feng, Y.Y., Ong, S.L., Hu, J.Y., Tan, X.L., Ng, W.J., 2003. Effects of
pH and temperature on the survival of coliphages MS2 and Q
beta. J. Ind. Microbiol. Biotechnol. 30 (9), 549e552.
Flint, J.A., Van Duynhoven, Y.T., Angulo, F.J., DeLong, S.M.,
Braun, P., Kirk, M., Scallan, E., Fitzgerald, M., Adak, G.K.,
Sockett, P., Ellis, A., Hall, G., Gargouri, N., Walke, H., Braam, P.,
2005. Estimating the burden of acute gastroenteritis,
foodborne disease, and pathogens commonly transmitted by
food: an international review. Clin. Infect. Dis. 41 (5), 698e704.
Fraisse, A., Temmam, S., Deboosere, N., Guillier, L., Delobel, A.,
Maris, P., Vialette, M., Morin, T., Perelle, S., 2011. Comparison
of chlorine and peroxyacetic-based disinfectant to inactivate
Feline calicivirus, Murine norovirus and Hepatitis A virus on
lettuce. Int. J. Food Microbiol. 151 (1), 98e104.
Francy, D.S., Stelzer, E.A., Bushon, R.N., Brady, A.M.G.,
Williston, A.G., Riddell, K.R., Borchardt, M.A., Spencer, S.K.,
Gellner, T.M., 2012. Comparative effectiveness of membrane
bioreactors, conventional secondary treatment, and chlorine
and UV disinfection to remove microorganisms from
municipal wastewaters. Water Res. 46 (13), 4164e4178.
Frederick, G.L., Lloyd, B.J., 1995. Evaluation by Serratia marcescenes
bacteriophage as a tracer and a model for virus removal in
waste stabilization ponds. Water Sci. Technol. 31 (12),
291e302.
Fumian, T.M., Leite, J.P.G., Castello, A.A., Gaggero, A., de
Caillou, M.S.L., Miagostovich, M.P., 2010. Detection of
rotavirus A in sewage samples using multiplex qPCR and an
evaluation of the ultracentrifugation and adsorptioneelution
methods for virus concentration. J. Virol. Methods 170 (1e2),
42e46.
Gandhi, K.M., Mandrell, R.E., Tian, P., 2010. Binding of virus-like
particles of Norwalk virus to romaine lettuce veins. Appl.
Environ. Microbiol. 76 (24), 7997e8003.
Gehr, R., Wagner, M., Veerasubramanian, P., Payment, P., 2003.
Disinfection efficiency of peracetic acid, UV and ozone after
enhanced primary treatment of municipal wastewater. Water
Res. 37 (19), 4573e4586.
Greer, A.L., Drews, S.J., Fisman, D.N., 2009. Why “winter” vomiting
disease? Seasonality, hydrology, and norovirus epidemiology
in Toronto, Canada. EcoHealth 6 (2), 192e199.
GVW, 2004. Shepparton Wastewater Management Facility.
Goulburn Valley Water, Inc, Shepparton, Victoria.
GVW, 2012. Water Supply Demand Strategy 2012e2060. Goulburn
Valley Water, Shepparton, Victoria.
GVW, 2013. Daily Raw Sewage Influent to Shepparton WMF (ML)
2012e13. Goulburn Valley Water, Shepparton, Australia.
Haagsma, J.A., Havelaar, A.H., Janssen, B.M., Bonsel, G.J., 2008.
Disability adjusted life years and minimal disease: application
of a preference-based relevance criterion to rank enteric
pathogens. Popul. Health Metr 6 (7), 1e8.
Haas, C., Rose, J., Gerba, C., 1999. Quantitative Microbial Risk
Assessment. John Wiley & Sons, New York.
Hall, G., Kirk, M., Ashbolt, R., Gregory, J., Lalor, K., Millard, G.,
Raupach, J., Sarna, M., Stafford, R., 2004. Results from the
National Gastroenteritis Survey 2001e2002. National Centre
for Epidemiology and Population Health, Australian National
University, Canberra, Australia.
Hall, G., Kirk, M.D., Becker, N., Gregory, J.E., Unicomb, L.,
Millard, G., Stafford, R., Lalor, K., 2005. Estimating foodborne
gastroenteritis, Australia. Emerg. Infect. Dis. 11 (8), 1257e1264.
Hall, G.V., Kirk, M.D., Ashbolt, R., Stafford, R., Lalor, K., OzFoodNet
Working, G., 2006. Frequency of infectious gastrointestinal
illness in Australia, 2002: regional, seasonal and demographic
variation. Epidemiol. Infect. 134 (1), 111e118.
Hamilton, A.J., Stagnitti, F., Premier, R., Boland, A.-M., Hale, G.,
2006. Quantitative microbial risk assessment models for
consumption of raw vegetables irrigated with reclaimed
water. Appl. Environ. Microbiol. 72 (5), 3284e3290.
Hamilton, A.J., Stagnitti, F., Xiong, X., Kreidl, S.L., Benke, K.K.,
Maher, P., 2007. Wastewater irrigation: the state of play.
Vadose Zone J. 6 (4), 823e840.
Harakeh, M.S., Butler, M., 1984. Factors influencing the ozone
inactivation of enteric viruses in effluent. Ozone: Sci. Eng. 6
(4), 235e243.
Haramoto, E., Katayama, H., Oguma, K., Yamashita, H.,
Tajima, A., Nakajima, H., Ohgaki, S., 2006. Seasonal profiles of
human noroviruses and indicator bacteria in a wastewater
treatment plant in Tokyo, Japan. Water Sci. Technol. 54
(11e12), 301e308.
Havelaar, A.H., Haagsma, J.A., Mangen, M.J.J., Kemmeren, J.M.,
Verhoef, L.P.B., Vijgen, S.M.C., Wilson, M., Friesema, I.H.M.,
Kortbeek, L.M., van Duynhoven, Y., van Pelt, W., 2012. Disease
burden of foodborne pathogens in the Netherlands, 2009. Int.
J. Food Microbiol. 156 (3), 231e238.
water research 54 (2014) 347e362 359
Hijnen, W.A.M., Beerendonk, E.F., Medema, G.J., 2006. Inactivation
credit of UV radiation for viruses, bacteria and protozoan (oo)
cysts in water: a review. Water Res. 40 (1), 3e22.
Joret, J.C., Block, J.C., Hartemann, P., Richard, Y., 1982.
Wastewater disinfection eelimination of fecal bacteria and
enteric viruses by ozone. Ozone Sci. Eng. 4 (2), 91e99.
Karavarsamis, N., Hamilton, A.J., 2010. Estimators of annual
probability of infection for quantitative microbial risk
assessment. J. Water Health 8 (2), 365e373.
Katayama, H., Haramoto, E., Oguma, K., Yamashita, H.,
Tajima, A., Nakajima, H., Ohyaki, S., 2008. One-year monthly
quantitative survey of noroviruses, enteroviruses, and
adenoviruses in wastewater collected from six plants in Japan.
Water Res. 42 (6e7), 1441e1448.
Kemmeren, J.M., Mangen, M.-J.J., van Duynhoven, Y.T.H.P.,
Havelaar, A.H., 2006. Priority Setting of Foodborne Pathogens:
Disease Burden and Costs of Selected Enteric Pathogens.
National Institute for Public Health and the Environment,
Bilthoven, The Netherlands.
Kukavica-Ibrulj, I., Darveau, A., Jean, J., Fliss, I., 2004. Hepatitis A
virus attachment to agri-food surfaces using immunological,
virological and thermodynamic assays. J. Appl. Microbiol. 97
(5), 923e934.
Kukkula, M., Maunula, L., Silvennoinen, E., von Bonsdorff, C.H.,
1999. Outbreak of viral gastroenteritis due to drinking water
contaminated by Norwalk-like viruses. J. Infect. Dis. 180 (6),
1771e1776.
La Rosa, G., Pourshaban, M., Iaconelli, M., Muscillo, M., 2010.
Quantitative real-time PCR of enteric viruses in influent and
effluent samples from wastewater treatment plants in Italy.
Ann. Ist. Super. Sanita 46 (3), 266e273.
Lake, R.J., Cressey, P.J., Campbell, D.M., Oakley, E., 2010. Risk
ranking for foodborne microbial hazards in New Zealand:
burden of disease estimates. Risk Anal. 30 (5), 743e752.
Langlet, J., Gaboriaud, F., Gantzer, C., Duval, J.F.L., 2008. Impact of
chemical and structural anisotropy on the electrophoretic
mobility of spherical soft multilayer particles: the case of
bacteriophage MS2. Biophys. J. 94 (8), 3293e3312.
Laverick, M.A., Wyn-Jones, A.P., Carter, M.J., 2004. Quantitative RT-
PCR for the enumeration of noroviruses (Norwalk-like viruses)
in water and sewage. Lett. Appl. Microbiol. 39 (2), 127e136.
Lazarova, V., Savoye, P., Janex, M.L., Blatchley, E.R.,
Pommepuy, M., 1999. Advanced wastewater disinfection
technologies: state of the art and perspectives. Water Sci.
Technol. 40 (4e5), 203e213.
Le Pendu, J., Ruvoen-Clouet, N., Kindberg, E., Svensson, L., 2006.
Mendelian resistance to human norovirus infections. Semin.
Immunol. 18 (6), 375e386.
Limpert, E., Stehel, W., Abbt, M., 2001. Log-normal distribution
across the sciences: keys and clues. Bioscience 51 (5), 341e352.
Lindesmith, L., Moe, C., Marionneau, S., Ruvoen, N., Jiang, X.,
Lindbland, L., Stewart, P., LePendu, J., Baric, R., 2003. Human
susceptibility and resistance to Norwalk virus infection. Nat.
Med. 9 (5), 548e553.
Locas, A., Martinez, V., Payment, P., 2010. Removal of human
enteric viruses and indicator microorganisms from domestic
wastewater by aerated lagoons. Can. J. Microbiol. 56 (2),
188e194.
Lodder, W.J., Husman, A.M.D., 2005. Presence of noroviruses and
other enteric viruses in sewage and surface waters in The
Netherlands. Appl. Environ. Microbiol. 71 (3), 1453e1461.
Lopman, B.A., Reacher, M.H., Vipond, I.B., Hill, D., Perry, C.,
Halladay, T., Brown, D.W., Edmunds, W.J., Sarangi, J., 2004.
Epidemiology and cost of nosocomial gastroenteritis, Avon,
England, 2002e2003. Emerg. Infect. Dis. 10 (10), 1827e1834.
Lucena, F., Araujo, R., Jofre, J., 1996. Usefulness of bacteriophages
infecting Bacteroides fragilis as index microorganisms of
remote faecal pollution. Water Res. 30 (11), 2812e2816.
Ma, J., Betts, N.M., Horacek, T., Georgiou, C., White, A., 2003.
Assessing stages of change for fruit and vegetable intake in
young adults: a combination of traditional staging algorithms
and food-frequency questionnaires. Health Educ. Res. 18 (2),
224e236.
Mara, D., Sleigh, A., 2010a. Estimation of norovirus and Ascaris
infection risks to urban farmers in developing countries using
wastewater for crop irrigation. J. Water Health 8 (3), 572e576.
Mara, D., Sleigh, A., 2010b. Estimation of norovirus infection risks
to consumers of wastewater-irrigated food crops eaten raw. J.
Water Health 8 (1), 39e43.
Matthews, J.E., Dickey, B.W., Miller, R.D., Felzer, J.R., Dawson, B.P.,
Lee, A.S., Rocks, J.J., Kiel, J., Montes, J.S., Moe, C.L.,
Eisenberg, J.N.S., Leon, J.S., 2012. The epidemiology of
published norovirus outbreaks: a review of risk factors
associated with attack rate and genogroup. Epidemiol. Infect.
140 (7), 1161e1172.
Maynard, H.E., Ouki, S.K., Williams, S.C., 1999. Tertiary lagoons: a
review of removal mechanisms and performance. Water Res.
33 (1), 1e13.
Mitakakis, T.Z., Sinclair, M.I., Fairley, C.K., Lightbody, P.K.,
Leder, K., Hellard, M.E., 2004. Food safety in family homes in
Melbourne, Australia. J. Food Prot. 67 (4), 818e822.
Mok, H.-F., Hamilton, A.J., 2014. Exposure factors for wastewater-
irrigated Asian vegetables and a probabilistic rotavirus disease
burden model for their consumption. Risk Anal. http://
dx.doi.org/10.1111/risa.12178.
Murray-Darling Basin Authority, 2012. Basin Plan. Murray-Darling
Basin Authority, Commonwealth of Australia, Canberra,
Australia.
Ni Mhurchu, C., Rodgers, A., Pan, W.H., Gu, D.F., Woodward, M.,
Parag, V., Lin, R., Bennett, D.A., Vander Hoorn, S., Barzi, F.,
MacMahon, S., Lam, T.H., Lawes, C., Suh, I., Ueshima, H.,
Okayama, A., Maegawa, H., Aoki, N., Nakamura, M., Kubo, N.,
Yamada, T., Wu, Z.S., Yao, C.H., Liu, L.S., Xie, J.X.,
Knuiman, M.W., Christensen, H., Wu, X.G., Zhou, J., Yu, X.H.,
Tamakoshi, A., Wu, Z.L., Chen, L.Q., Shan, G.L., Duan, X.F.,
Norton, R., Whitlock, G., Jackson, R., Fujishima, M.,
Kiyohara, Y., Iwamoto, H., Woo, J., Ho, S., Hong, Z.,
Huang, M.S., Zhou, B., Fuh, J.L., Kita, Y., Choudhury, S.R.,
Jee, S.H., Kim, I.S., Giles, G., Hashimoto, T., Sakata, K., Imai, Y.,
Ohkubo, T., Hozawa, A., Jamrozik, K., Hobbs, M.,
Broadhurst, R., Nakachi, K., Fang, X.H., Li, S.C., Yang, Q.D.,
Chen, Z.M., Tanaka, H., Nozaki, A., Horibe, H., Matsutani, Y.,
Kagaya, M., Hughes, K., Lee, J., Heng, D., Zhou, B.F.,
Zhang, H.Y., Shimamoto, K., Saitoh, S., Li, Z.Z., He, Y.,
Yao, S.X., 2004. Body mass index and cardiovascular disease in
the Asia-Pacific region: an overview of 33 cohorts involving
310 000 participants. Int. J. Epidemiol. 33 (4), 751e758.
Nordgren, J., Matussek, A., Mattsson, A., Svensson, L.,
Lindgren, P.-E., 2009. Prevalence of norovirus and factors
influencing virus concentrations during one year in a full-scale
wastewater treatment plant. Water Res. 43 (4), 1117e1125.
NRMMC, EPHC, AHMC, 2006. Australian guidelines for water
recycling: managing health and environmental risks. In:
Natural Resource Management Ministerial Council,
Environment Protection and Heritage Council, and the
Australian Health Ministers’ Conference Canberra, Australia.
Oppenheimer, J.A., Jacangelo, J.G., Laı
ˆne
´, J.M., Hoagland, J.E., 1997.
Testing the equivalency of ultraviolet light and chlorine for
disinfection of wastewater to reclamation standards. Water
Environ. Res. 69 (1), 14e24.
Oragui, J.I., Arridge, H., Mara, D.D., Pearson, H.W., Silva, S.A., 1995.
Rotavirus removal in experimental waste stabilization pond
systems with different geometries and configurations. Water
Sci. Technol. 31 (12), 285e290.
Oragui, J.I., Curtis, T.P., Silva, S.A., Mara, D.D., 1987. The removal
of excreted bacteria and viruses in deep waste stabilization
water research 54 (2014) 347e362360
ponds in northeast Brazil. Water Sci. Technol. 19 (3e4),
569e573.
Ottoson, J., Stenstrom, T.A., 2003. Faecal contamination of
greywater and associated microbial risks. Water Res. 37 (3),
645e655.
Persson, J., Wittgren, H.B., 2003. How hydrological and hydraulic
conditions affect performance of ponds. Ecol. Eng. 21 (4e5),
259e269.
Petterson, S.R., Ashbolt, N.J., 2001. Viral risks associated with
wastewater reuse: modeling virus persistence on wastewater
irrigated salad crops. Water Sci. Technol. 43 (12), 23e26.
Petterson, S.R., Ashbolt, N.J., Sharma, A., 2001. Microbial risks
from wastewater irrigation of salad crops: a screening-level
risk assessment. Water Environ. Res. 73 (6), 667e672.
Petterson, S.R., Ashbolt, N.J., Sharma, A., 2002. Of: microbial risks
from wastewater irrigation of salad crops: a screening-level
risk assessment. Water Environ. Res. 74 (4), 411e421.
Plum, V., Dahl, C.P., Bentsen, L., Petersen, C.R., Napstjert, L.,
Thomsen, N.B., 1998. The actiflo method. Water Sci. Technol.
37 (1), 269e275.
Radcliffe, J.C., 2010. Evolution of water recycling in Australian
cities since 2003. Water Sci. Technol. 62 (4), 792e802.
RIRDC, 2011. Taking Stock of the Australian Asian Vegetables
Industry. Autsralian Rural Industries Research and
Development Corporation and Horticulture Australia Limited,
NWS Australia.
RMCG, GSCC, 2012. Adaptation Action Plan for a Future with Less
Water. Greater Shepparton City Council and RMCG
Consultants for Business, Communities, and Environment,
Shepparton, Victoria.
Rose, J.B., Dickson, L.J., Farrah, S.R., Carnahan, R.P., 1996. Removal
of pathogenic and indicator microorganisms by a full-scale
water reclamation facility. Water Res. 30 (11), 2785e2797.
Rutjes, S.A., Italiaander, R., van den Berg, H., Lodder, W.J.,
Husman, A.M.D., 2005. Isolation and detection of enterovirus
RNA from large-volume water samples by using the NucliSens
miniMAG system and real-time nucleic acid sequence-based
amplification. Appl. Environ. Microbiol. 71 (7), 3734e3740.
Rydell, G.E., Kindberg, E., Larson, G., Svensson, L., 2011.
Susceptibility to winter vomiting disease: a sweet matter. Rev.
Med. Virol. 21 (6), 370e382.
Saqqar, M.M., Pescod, M.B., 1995. Modeling the performance of
anaerobic wastewater stabilization ponds. Water Sci. Technol.
31 (12), 171e183.
Scherrenberg, S.M., 2006. Treatment Techniques for Combined
Sewer Overflows: A Literature Study of the Available
Techniques. Delft University of Technology, Delft,
Netherlands.
Schoen, M.E., Soller, J.A., Ashbolt, N.J., 2011. Evaluating the
importance of faecal sources in human-impacted waters.
Water Res. 45 (8), 2670e2680.
Seidu, R., Heistad, A., Amoah, P., Drechsel, P., Jenssen, P.D.,
Stenstrom, T.-A., 2008. Quantification of the health risk
associated with wastewater reuse in Accra, Ghana: a
contribution toward local guidelines. J. Water Health 6 (4),
461e471.
Shapiro, O.H., Kushmaro, A., Brenner, A., 2010. Bacteriophage
predation regulates microbial abundance and diversity in a
full-scale bioreactor treating industrial wastewater. ISME J. 4
(3), 327e336.
Shuval, H., Lampert, Y., Fattal, B., 1997. Development of a risk
assessment approach for evaluating wastewater reuse
standards for agriculture. Water Sci. Technol. 35 (11e12),
15e20.
Sigmund, T., Siczka, J., Elliott, T., Awad, J., Onderko, R., 2006.
Operating chemically enhanced clarification for optimum
disinfection performance. In: Proceedings of the Water
Environment Federation, WEFTEC 2006, pp. 6707e6722.
Silverman, A.I., Nelson, K.L., Akrong, M.O., Amoah, P.,
Drechsel, P., 2013. Quantification of human norovirus GII,
human adenovirus, and fecal indicator organisms in
wastewater used for irrigation in Accra, Ghana. J. Water
Health 11 (3), 473e488.
Sinclair, M.I., Hellard, M.E., Wolfe, R., Mitakakis, T.Z., Leder, K.,
Fairley, C.K., 2005. Pathogens causing community
gastroenteritis in Australia. J. Gastroenterol. Hepatol. 20 (11),
1685e1690.
Sinton, L.W., Hall, C.H., Lynch, P.A., Davies-Colley, R.J., 2002.
Sunlight inactivation of fecal indicator bacteria and
bacteriophages from waste stabilization pond effluent in
fresh and saline waters. Appl. Environ. Microbiol. 68 (3),
1122e1131.
Sobsey, M.D., 1989. Inactivation of health-related microorganisms
in water by disinfection processes. Water Sci. Technol. 21 (3),
179e195.
Soller, J.A., Schoen, M.E., Bartrand, T., Ravenscroft, J.E.,
Ashbolt, N.J., 2010. Estimated human health risks from
exposure to recreational waters impacted by human and non-
human sources of faecal contamination. Water Res. 44 (16),
4674e4691.
Stine, S.W., Song, I., Choi, C.Y., Gerba, C.P., 2005. Effect of relative
humidity on preharvest survival of bacterial and viral
pathogens on the surface of cantaloupe, lettuce, and bell
peppers. J. Food Prot. 68 (7), 1352e1358.
Szal, G.M., Nolan, P.M., Kennedy, L.E., Barr, C.P., Bilger, M.D., 1991.
The toxicity of chlorinated wastewater: Instream and
laboratory case studies. Res. J. Water Pollut. Control Fed. 63 (6),
910e919.
ter Waarbeek, H.L.G., Dukers-Muijrers, N., Vennema, H.,
Hoebe, C., 2010. Waterborne gastroenteritis outbreak at a
scouting camp caused by two norovirus genogroups: GI and
GII. J. Clin. Virol. 47 (3), 268e272.
Teunis, P.F.M., Moe, C.L., Liu, P., Miller, S.E., Lindesmith, L.,
Baric, R.S., Le Pendu, J., Calderon, R.L., 2008. Norwalk virus:
how infectious is it? J. Med. Virol. 80 (8), 1468e1476.
The R Foundation for Statistical Computing, 2013. The R Project
for Statistical Computing.
Thebault, A., Teunis, P.F.M., Le Pendu, J., Le Guyader, F.S., Denis, J.-
B., 2013. Infectivity of GI and GII noroviruses established from
oyster related outbreaks. Epidemics 5 (2), 98e110.
Thorven, M., Grahn, A., Hedlund, K.O., Johansson, H.,
Wahlfrid, C., Larson, G., Svensson, L., 2005. A homozygous
nonsense mutation (428G/A) in the human secretor (FUT2)
gene provides resistance to symptomatic norovirus (GGII)
infections. J. Virol. 79 (24), 15351e15355.
Toze, S., 2006. Reuse of effluent water ebenefits and risks. Agric.
Water Manag. 80 (1e3), 147e159.
Tree, J.A., Adams, M.R., Lees, D.N., 2003. Chlorination of indicator
bacteria and viruses in primary sewage effluent. Appl.
Environ. Microbiol. 69 (4), 2038e2043.
Tree, J.A., Adams, M.R., Lees, D.N., 2005. Disinfection of Feline
calicivirus (a surrogate for norovirus) in wastewaters. J. Appl.
Microbiol. 98 (1), 155e162.
US EPA, 2003. CSFII Analysis of Food Intake Distributions. US
Environmental Protection Agency, Washington DC, USA.
US EPA, 2011. Exposure Factors Handbook, 2011 ed. Research and
Development, National Center for Environmental
Assessment, U.S. Environmental Protection Agency,
Washington DC.
van den Berg, H., Lodder, W., van der Poel, W., Vennema, H.,
Husman, A.M.D., 2005. Genetic diversity of noroviruses in raw
and treated sewage water. Res. Microbiol. 156 (4), 532e540.
Verhoef, L., Koopmans, M., Van Pelt, W., Duizer, E., Haagsma, J.,
Werber, D., Van Asten, L., Havelaar, A., 2013. The estimated
disease burden of norovirus in The Netherlands. Epidemiol.
Infect. 141 (3), 496e506.
water research 54 (2014) 347e362 361
Walls, H.L., Wolfe, R., Haby, M.M., Magliano, D.J., de Courten, M.,
Reid, C.M., McNeil, J.J., Shaw, J., Peeters, A., 2010. Trends in
BMI of urban Australian adults, 1980e2000. Public Health Nutr.
13 (5), 631e638.
Watts, R.J., Kong, S.H., Orr, M.P., Miller, G.C., Henry, B.E., 1995.
Photocatalytic inactivation of coliform bacteria and
viruses in secondary wastewater effluent. Water Res. 29 (1),
95e100.
Weisstein, E.W., 2013. Pfaff Transformation. Mathworld eA
Wolfram Web Resource, USA.
WHO, 2006. WHO guidelines for the safe use of wastewater,
excreta and greywater. In: Wastewater Use in Agriculture, vol.
II. World Health Organisation, Geneva, Switzerland.
Widdowson, M.A., Sulka, A., Bulens, S.N., Beard, R.S., Chaves, S.S.,
Hammond, R., Salehi, E.D.P., Swanson, E., Totaro, J., Woron, R.,
Mead, P.S., Bresee, J.S., Monroe, S.S., Glass, R.I., 2005.
Norovirus and foodborne disease, United States, 1991e2000.
Emerg. Infect. Dis. 11 (1), 95e102.
Wolfram Research, 2013. Hypergeometric 2F1. Wolfram Research,
Inc.
Wu, J., Long, S.C., Das, D., Dorner, S.M., 2011. Are microbial
indicators and pathogens correlated? A statistical analysis of
40 years of research. J. Water Health 9 (2), 265e278.
Xu, P., Janex, M.L., Savoye, P., Cockx, A., Lazarova, V., 2002.
Wastewater disinfection by ozone: main parameters for
process design. Water Res. 36 (4), 1043e1055.
water research 54 (2014) 347e362362
