Quantitative Microbial Risk Assessment for Recreational Waters at Three Lake Michigan Beaches

Quantitative Microbial Risk Assessment
for Recreational Waters at Three Lake
Michigan Beaches
Tucker Burch1,3, Steven R. Corsi1, Susan K.
Spencer2,3, Rebecca B. Carvin1, Mark A.
Borchardt2,3
1USGS,
Wisconsin Water Science Center
2USDA – Agricultural Research Service
3Laboratory for Infectious Disease and the Environment
U.S. Department of the Interior
U.S. Geological Survey
This information is preliminary and is subject to revision. It is being provided to meet the
need for timely best science. The information is provided on the condition that neither the
U.S. Geological Survey nor the U.S. Government shall be held liable for any damages
resulting from the authorized or unauthorized use of the information.
Pathogen exposure in recreational water
Artwork: Ben Siebers (USGS WI WSC)
Why QMRA?
Policy Makers and Resource Managers
Human
activities
Abundance
Pathogens
Environmental
conditions
Distribution
Health risk
• Human
• Livestock
• Wildlife
QMRA, in a nutshell
Exposure
Assessment
Pathogen
Concentration:
C = 5 Salmonella/L
Estimate Risk
(Monte Carlo)
DoseResponse
Model
Risk:
P(response) = 50%
Probability of Response
Probability of Response
Volume Ingested:
V = 10 mL
Exposure:
Dose = C×V
Dose-Response
Assessment
Dose
Risk
Dose
Questions



What’s the risk relative to recreational water quality
criteria (RWQC)?
What factors can be manipulated to mitigate risk?
What do we need to know to more fully characterize
risk?
?
Approach
1.
2.
3.
Quantify pathogens at beaches of interest
Develop pathogen models as functions of
environmental predictors (e.g., water
temperature, cloud cover)
Estimate risk from modeled concentrations
Watersheds and beaches
 Clay Banks


63% forested wetland, 17%
pasture/hay
WWTP effluent, impervious runoff,
agricultural runoff, septic systems
 Point Beach

30% pasture/hay, 28% crops, 28%
forested wetland
 Red Arrow, 2 subwatersheds


Storm sewer (100% urban)
Manitowoc River (70% agricultural)
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Methods


Sampling
 ~20 times per beach, Memorial Day to Labor Day, 2010
 Glass wool filtration
Analyses
 qPCR for 21 human, bovine, and zoonotic pathogens
 Culture for two bacterial pathogens in QMRA
(Campylobacter jejuni and Salmonella spp.)
 Pathogen modeling
 Tobit (censored) regression models
 Screened ~200 candidate predictors
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Pathogen abundance and distribution
Three most prevalent pathogens:
Enteroviruses
C. jejuni
Salmonella spp.
Points = median concentration
Whiskers = 25th and 75th percentiles
ND = non-detect
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Model results

Enterovirus predictors:
1. Maximum 1-hour parallel current
2. Mean 12-hour long-shore wave component
3. Mean 12-hour off-shore wave component
4. 10-day water temperature

C. jejuni predictors:
1. Maximum 12-hour cloud cover
2. Beach (Point)

Salmonella spp. predictors:
 None!
Preliminary Information-Subject to Revision. Not for Citation or Distribution
QMRA methods
Exposure
Assessment
•
Enteroviruses: echovirus,
feeding study-based
(CAMRA 2013)
C. jejuni: feeding studybased (Schmidt et al. 2013)
Salmonella: outbreakbased (Teunis et al. 2010)
•
•
Estimate Risk
(Monte Carlo)
•
Two-dimensional: variability
and uncertainty
10,000 simulations in
variability dimension
1,000 simulations in
uncertainty dimension
•
•
Probability of Response
•
•
Enteroviruses and C. jejuni:
Tobit models
Salmonella: raw data
Ingestion rates and swim
times derived from Suppes
et al. 2014
Probability of Response
•
Dose-Response
Assessment
Dose
Risk
Dose
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Salmonella spp. simulation inputs
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Risk estimates
Median = 3×10-7
Median = 4×10-5
Median = 8×10-6
Solid lines = median in uncertainty dimension
Dashed lines = 2.5th, 25th, 75th, and 97.5th percentiles in uncertainty dimension
Vertical dashed line = EPA recreational water quality risk benchmark (32 illnesses per 1,000
recreators per event)
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Influence of simulated risk factors
Bars = Median of Spearman’s (rank order) correlation coefficient between risk estimates and
simulation inputs (on vertical axis)
Whiskers = 2.5th and 97.5th percentiles of Spearman’s correlation coefficient between risk
estimates and simulation inputs
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Questions

What’s the risk relative to RWQC?
 Median risk < RWQ benchmark (32 illnesses per 1,000



recreators) for all 3 pathogens
6.4% of Salmonella risk estimates exceed benchmark, vs.
0.2% and 0% for C. jejuni and enteroviruses
What factors can be manipulated to mitigate risk?
 For enteroviruses and C. jejuni: ingestion rate?
 For Salmonella: concentration
What do we need to know to more fully characterize
risk?
 Factors controlling Salmonella concentrations
 Salmonella serotypes
?
Preliminary Information-Subject to Revision. Not for Citation or Distribution
Acknowledgements
 Collaborators
 Colleen McDermott (UW-Oshkosh)
 Greg Kleinheinz (UW-Oshkosh)
 Austin Baldwin (USGS-WI WSC)

Funding
 Great Lakes Restoration Initiative
 Ocean Research Priorities Plan
References
Center for Advancing Microbial Risk Assessment (CAMRA). 2013. Echovirus: Dose
response models. QMRAwiki, accessed May 11, 2015,
http://qmrawiki.canr.msu.edu/index.php/Echovirus:_Dose_Response_Models#_4ac
d021e7e230054919792f80365b5cd.
Schmidt PJ, Pintar KDM, Fazil AM, Topp E. 2013. Harnessing the theoretical
foundations of the exponential and beta-Poisson dose-response models to quantify
parameter uncertainty using Markov Chain Monte Carlo. Risk Analysis, 33(9):16771693.
Suppes LM, Abrell L, Dufour AP, Reynolds KA. 2014. Assessment of swimmer
behaviors on pool water ingestion. Journal of Water and Health, 12(2):269-279.
Teunis PFM, Kasuga F, Fazil A, Ogden ID, Rotariu O, Strachan NJC. 2010. Doseresponse modeling of Salmonella using outbreak data. International Journal of
Food Microbiology, 144(2):243-249.