Utility of viral pathogen
testing as a regulatory
parameter for EU bathing
waters
Peter Wynn-Jones, Jane Sellwood, Mark Wyer and David Kay
Policy
Context
Projects
Virobathe (FP6)
Epibathe (FP6)
Viroclime (FP7)
Activites 1. Methods development (Virobathe)
Filters vs glass-wool and bovril vs marvel
2. Inter-laboratory AQC
3. Correlations with health-related FIOs
4. Acqusition of surveillance data
1. Fresh
2. Marine
3. Sewage
1. Raw
2. Secondary
3. tertiary
5. Assessment as a regulatory parameter
Viroclime
Adenovirus results grouped by quartiles of Escherichia coli
concentration - marine water
p = significance of Mantel-Haenszel linear trend
test not significant at the 95% confidence level (p > 0.05)
* the colour shading
indicates the proportion
of results in each
category
Adenovirus results grouped by quartiles of Intestinal
enterococci concentration - fresh water
p = significance of Mantel-Haenszel linear trend
† test significant at the 95% confidence level (p < 0.05)
* the colour shading
indicates the proportion
of results in each
category
Categorical analysis of Adenovirus results - Summary, quartile
interval, conservative
p = significance of Mantel-Haenszel linear trend
† test significant at the 95% confidence level (p < 0.05)
Parameter Type p
Escherichia coli Fresh †0.000
Escherichia coli Marine 0.164
Intestinal enterococci Fresh †0.000
Intestinal enterococci Marine 0.793
Somatic coliphage Fresh †0.000
Somatic coliphage Marine 0.461
Norovirus GGII Analysis
928 Fresh water analyses (232 “tetrads”)
482 Marine water analyses (121 “tetrads”)
Conservative - PCR fail = 0
Affects 5.0% of fresh water results
and 7.9% of marine results
Categorical analysis of Norovirus GGII results
n
Fresh - 232
Marine - 121
* the colour shading
indicates the proportion
of results in each
category
Italic figures at the base of the bars =
number of positive “tetrads”
Norovirus GGII results grouped by quartiles of Intestinal
enterococci concentration - marine water
p = significance of Mantel-Haenszel linear trend
test not significant at the 95% confidence level (p > 0.05)
* the colour shading
indicates the proportion
of results in each
category
Categorical analysis of Norovirus GGII results - Summary,
quartile interval, conservative
p = significance of Mantel-Haenszel linear trend
† test significant at the 95% confidence level (p < 0.05)
Parameter Type p
Escherichia coli Fresh †0.002
Escherichia coli Marine 0.050
Intestinal enterococci Fresh 0.054
Intestinal enterococci Marine 0.361
Somatic coliphage Fresh †0.000
Somatic coliphage Marine 0.106
Ring Trials Can two labs give comparable answers on the same
sample?
Results of Mineral Water Matrix Ring Trial
May 2011
UMU and FIOCRUZ
have mean values
which are not
statistically different
Results of Sewage Matrix Ring Trial
May 2011
All labs have mean
values which are
statistically different
Advice Lack of correlation with known FIOs
Testing by qPCR was not reproducible with the
precision needed for regulatory parameter
Wide swings (peaks) in key pathogens such as
Norovirus
Indicates pathogens shedders in the population NOT
‘risk’ in a precautionary assessment
Needs more epidemiology
Swansea Bay Model
Validation
November 2015
Mark Wyer, David Kay and the SCSC Team
Swansea Bay model NRW Hydrometry
inputs
SCSC Installation
Inputs
Calculated Variable Inputs
Black Pill – UVA,
wind speed
Clyne River -
discharge
Afan - discharge
Dulais - discharge
Workbook
Output
Sign:
Good
Poor
Without discounting = SUFFICIENT
Discounting using the model/signage:
2013 – 2 samples discounted/0 replaced
2014 – 2 samples discounted/2 replaced
2015 – 1 sample discounted/1 replaced
Official classification = GOOD
A clear demonstration of the utility of applying
prediction and discounting under Directive 2006/7/EC
2012-2015 Classification
DSP Intensive sampling
Pronounced
variation within
the sampling day
Water quality - pGI Thresholds
Water quality thresholds relating to pGI can be
developed using relationships between Mean log10
enterococci and pGI
Asymptotic sigmoid
functions –
Richard’s Curve –
fitted to data
Site specific model
for Swansea DSP
2014 validation data
20 days
1 day in each week of the bathing season
15/05/2014 – 30/09/2014
30-minute sampling between 08:00 and 17:00 BST
19 samples per day
Sampled as per 2011
Enterococci analysis using triplicate filtrations to
improve precision – as per 2011
Results – Individual samples
Familiar pattern of variation in IE concentration within each
sampling day
Results – Daily GM
Variation in daily geometric mean IE concentration:
the measure predicted by the model
Comparison of daily results - GM
Comparison with 2011 results – statistically
similar
Comparison of daily results - SD
Comparison with 2011 results – No statistically
significant change
Diurnality – AM vs PM
GM IE concentration was lower in the afternoon
Difference was statistically significant overall
and when daily pGI > 0.1
2014 model fit
Model over predicts in late September?
Within-season decline in solar radiation – near equinox?
Problems with UVA sensor shading?
2014 model fit
8 points above
the line (P=0.4)
12 points below
the line (P=0.6)
Binomial test for
Ho: P=0.5
p=0.5034
Model fit: 2011 & 2014 data
2011:
29 points above
the line (P=0.47)
30 points below
the line (P=0.51)
Binomial test for
Ho: P=0.5
p=1.0000
Performance - Misclassification
2011
2014
Critical misclassification:
≈ 5%
Thanks
City and County of Swansea Pollution Control Division
staff
Natural Resources Wales Analytical Services staff
Patrick Lilly – Welsh European Funding Office
Strathclyde Loch Microbial
Tracer Study
November 2015
Mark Wyer, David Kay and the CREH team
Background
Strathclyde Loch was the venue for the prestigious triathlon events in the Glasgow 2014 Commonwealth Games
An illness outbreak, attributed to norovirus, was recorded following a competitive event in summer 2012 - reported by John Logan at the Southport Symposium
A curtain barrier system was deployed to partition a swim area for the events
A microbial tracer experiment was designed to test the effectiveness of the barrier against virus particles
Barrier curtain system
Tracer organism
Large plaque phage isolated from sea water
Infects Serratia marcescens bacteria
Plaques – clear areas
where the phage has
infected the host bacteria
Units: plaque forming
units (pfu)/ml
Study design
Prior sampling showed absence of the tracer organism
in the loch
Continuous hourly sampling at 9 locations around the
loch for 72 hours – 11th – 14th November 2013
1 litre of tracer released where South Calder Water
enters the loch
1.0 x 1016 particles of S. marcescens bacteriophage
released
Nine sampling sites
Sampling
Measures taken to avoid potential cross contamination
by the sampling team itself
A system of completely independent sampling
implemented for the sites in the two areas
Two sets of dedicated sampling equipment – clearly
marked and cleaned after each sample
Samples from inside and outside the swim zone were
stored and transported in separate cool boxes
Outside swim zone - south
Outside barrier at pontoons
Tracer “breakthrough” 23 h after release
Inside swim zone - south
Inside barrier near causeway to tower
Tracer present in one sample (2 pfu/ml) 31 h post release
Suggests the curtain barrier is very effective at this site
Swim zone – south east
Inside barrier adjacent to the erstwhile overflow to the Clyde
Tracer present in two samples (1 pfu/ml) 44 and 63 h post release
- Suggests the curtain barrier is very effective here
Outside swim zone - north
Outside barrier near rowing course finish marker
Tracer “breakthrough” 18 h post release
Inside swim zone - north
Inside barrier at rowing course finish marker
Tracer “breakthrough” 25 h post release – leakage past barrier
Indicates that the barrier not as effective at this site
Reasons
Incomplete sealing of the barrier at the shore – physical
connection observed at the start of the study
Blow over from waves generated in the larger area of
the loch was observed during sampling
Closure of the two sides of the barrier observed during
the study
Suggests the barrier requires management at this side
of the loch to be fully effective
Tracer transfer by loch waterfowl?
Thanks
Ian Spiers – Scottish Government
Calum McPhail - SEPA
Brian Devlin – North Lanarkshire Council
Sandyhills Bay Tracer and
Water Quality Study
November 2015
Mark Wyer, David Kay and the CREH team
Study Design
Release of 4 microbial tracer organisms to inputs discharging to Sandyhills Bay – bacteriophage
5 litres of tracer at each site
MS2 coliphage – 5 x 1017 particles
Serratia marcescens phage – 4.3 x 1016 particles
fX174 bacteriophage – 1.45 x 1017 particles
Enterobacter cloacae phage – 6 x 1015 particles
Continuous hourly sampling at at the beach for 54 hours: 22nd – 24th September 2013
Analysis for faecal indicator organisms, turbidity and salinity
Sampling
Wide tide range and rapid tidal movement required use
of an ATV on the beach to cover all states
Tracer release and sampling
Additional sediment sampling
Results – MS2 Barnhourie Burn
Tracer appears 4-5 h after release
Present in most subsequent samples
Elevation on alternating tides?
Results – S. marcescens phage Sandyhills Car Park Septic Tank
Tracer appears 5 h after release
Present in most subsequent samples
Strong signal on the next incoming, remaining through low water
Results – fX174 phage
Powbrade Burn
Tracer appears 9 h after release
Present in all subsequent samples
Pattern appears to relate to tide – higher concentration at HW
Results – Ent. cloacae phage Southwick Water
Tracer appears 12 h after release
Present in all subsequent samples
Initial peak at next HW, gradually decaying on subsequent tides
Results – phage tracers in sediment
MS2 and S. marcescens phage from Barnhourie Burn present in
samples from both low tides
Higher MS2 reflects higher tracer titre
Results – phage tracers in sediment
Ent. cloacae phage from Southwick Water not present in samples
from from the first tide – consistent with the tracer time series
fX174 from Powbrade Burn present on both tides
Results – FIOs Day vs. Night
No statistically significant differences (Student’s t-test p > 0.05)
Results – FIOs inside/outside bay
Statistically significant differences (Student’s t-test p < 0.05)
GM FIO concentrations were higher inside the bay
Results – FIOs in sediment
FIOs were present in sediments – higher levels of enterococci
(104cfu/100g) cf. E. coli (5x103 cfu/100g)
Difference in survival characteristics?
Results – Turbidity & salinity
Clear pattern of reduced salinity associated with the incoming
tide and through HW – freshwater trapped in the smaller bay by
the incoming tide?
Results – inside/outside bay
Mean salinity significant different inside the bay
(Student’s t-test p < 0.05)
No significant difference in GM turbidity
Conclusions
Tracers demonstrate connectivity between all 4 labelled
sources and the DSP through successive tidal cycles
Tracer patterns appear to reflect the tidal sequence
FIO concentrations show tidal rather than diurnal (i.e.
day night) relationships
Samples collected within the bay had significantly
higher FIO concentrations and significantly lower
salinity than those collected outside
Conclusions
The pattern suggests that water with a relatively high
freshwater component and FIO concentration from the
adjacent catchment is trapped in the smaller bay area
on incoming tides
Microbiological analysis of sediments showed presence
of FIOs
Tracer organisms – were also deposited in the
sediments – probably through adsorption to sediment
particles
Thanks
Ian Spiers – Scottish Government
Calum McPhail – SEPA
Ruth Stidson – SEPA
David Findlay – SEPA
Andy Thomas – Gillespie Leisure
Further information
Professor David Kay:
Dr Mark Wyer: