"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
Water Quality and GI Illness in Ecuador and Atlanta
1. Karen Levy, PhD, MPH
October 2013
Salúd Ambiental en Chile: Orientación para Futuras
Investigaciones
Planning for a GeoHealth Hub in Chile
Agua Potable y Enfermedades
Gastrointestinales en Esmeraldas,
Ecuador y Atlanta, GA
2. •Overview of contamination between
source and point of use
•Case Study from Esmeraldas, Ecuador
•Case Study from Atlanta, GA
Agua Potable y Enfermedades
Gastrointestinales en Esmeraldas,
Ecuador y Atlanta, GA
15. N Enterococci
CFU/100mL
E.coli
CFU/100mL
Source 59 227.9 227.1
(139.5 – 372.4) (144.6-356.9)
Household 105 103.7 113.4
(70.7-152.1) (80.8-159.0)
Control 105 80.8 83.8
(54.3-120.1) (57.8-121.5)
Contamination:
Source > Point of Use
Contaminación Observada
16. Paired log differences:
Natural attenuation = Source – Control
In-home attenuation = Source – Household
In-home recontamination = Household - Control
Enterococci E.coli
(GEE regressions adjust for clustering by paired household samples,
and to account for autocorrelation between sampling days) n=105
17. Findings
Reductions of indicator organisms between source and POU…
…followed by recontamination in ~50% of households
Levy et al. 2008. Environmental Health Perspectives
116(11): 1533-1540.
This suggests the need to consider both initial water
quality conditions and recontamination
24. Gastrointestinal Disease Data
• 41 hospitals in Metro Atlanta
• 4.4 million ED records
• 254,760 GI illness records
(based on ICD-9 codes)
• Non-injury visits = comparison
group
• Address + zip-code data for
majority of records
25. Water Utility Data
• Hydraulic Models
• Utility Coverage Areas
• Water Residence Time from plant to node
(estimated through simulations of water
flow through the distribution system)
26. Source: Besner et al. 2011 Water Research 45: 961-979
Water
Residence
Time
ER visits
27. Utility 1 Utility 2
WRT: short 6.8 /intermediate 22.0 / long 47.4 hrs
WRT: short 10.1 / intermediate 33.4 / long 74.4 hrs
WRT: short 5.9 / intermediate 18.5 / long 60.4 hrs
28. ControlVariables
(Census Data & ED records)
• Age
• Season
• Year
• Hospital
• Distance from zip code centroid to hospital
• Zip code median income
• Zip code percent minority
• Medicaid payment status
• Age*Medicaid
• Age*Distance to hospital
• Medicaid*Distance to hospital
29. OddsRatio
Based onTinker et al. 2009. JWH 7(2): 332-343
.85
.85
.85.95
.95
.951.05
1.05
1.051.15
1.15
1.150
0
0.5
.5
.51
1
11.5
1.5
1.52
2
20
0
0.5
.5
.51
1
11.5
1.5
1.52
2
2Utility 1
Utility 1
Utility 1Utility 2
Utility 2
Utility 2Zipcode WRT, All addresses, 28 hospitals
Zipcode WRT, All addresses, 28 hospitals
Zipcode WRT, All addresses, 28 hospitalscomparison_no
comparison_no
comparison_no
Water Residence Time
Short Intermediate Long Short Intermediate Long
OR = 1.05
(1.00-1.09)
OR = 1.05
(1.02-1.08)
30. Conclusions:
• People living in zip codes receiving water
with the longest residence time in the
distribution system may be at moderately
increased risk for GI illness
31. Follow-up Study to Evaluate
Microbial Water Quality in the
Distribution System
• Water Residence Time
• Main Breaks
• Pressure Fluctuations
37. Ultrafiltración para
Muestras de GranVolumen
Recycle
Pump
Sample
reservoir100 L
Ultrafilter
Air vent
Retentate
reservoir
Q Permeate (waste)
Flow
Flow
Restrictor
Dead-end
Flow meter
1L
2,900 mL/min
1
,
2
1,100 mL/min
3-way valve
38. Water
Sample!
Filtered Water!
(aka “filtrate” or “permeate”)!
(UF Membrane)!
Microbes!
Suspended solids!
Colloidal matter!
Large dissolved molecules!
Hollow Fiber Ultrafiltration Schematic
Division of Foodborne,Waterborne and
Environmental Diseases at the CDC in
Atlanta with Dr.Vincent Hill!
39. Microbiological Analysis
Membrane
Filtration
~400 ml
Concentrate + Back flush
PEG
0.1 mL 1 mL 10 mL
E. coli; Total coliforms; C. perfringens;
HPC; P. aeruginosa; A. hydrophila
60 to 100 mL
Actual volume filtered for each
Standard Methods for the Examination of Water and Wastewater; 20th Edition, 1999 American Public Health Association Publications
Sample Collection
AMS (90L) or Grab sample (2L)
Ultrafiltration
(AMS, 90L)
100 mL
10^3 10^4 10^5
25 ml actual volume pour plate each
Male-specific (MS2) and Somatic
Coliphage
Dr. Nick Ashbolt lab
Mycobacterium avium
complex (MAC),
Legionella +amoebae,
Cryptosporidium
40.
41. 1 2 3 4 5 6
0
0. 5
1. 0
1. 5
2. 0
E
.
c
o
l
i
C
F
U
/
1
0
0
m
l
Si t e
42. Acknowledgements
• Funding: EPA-STAR grant #R834250
(PIs Christine Moe & Paige Tolbert)
• Co-Authors: Mitch Klein, Stefanie Sarnat, Sarah
Tinker, Samina Panwhar, Alexandra Huttinger,
Jim Uber, Paige Tolbert, Christine Moe
• Other:Water Utility staff, Hospital staff
R834250
The content of this presentation is solely the responsibility of the grantee
and does not necessarily represent the official views of the US EPA
43. Co-‐Authors:
• Kara
Nelson
• Alan
Hubbard
• Joe
Eisenberg
Field
Assistance:
• Sarah
Bates
• Sangam
Tiwari
• Laura
McLaughlin
• Karina
Ponce
• Patricio
Bueno
• William
Cevallos
• Geovanny
Hurtado
• Maritza
Renteria
• Deni
Tenorio
• Jairo
Ayovi
• Mirta
Campaz
• Melecio
Quintero
• Owen
Solberg
• Khalid
Kadir
ACKNOWLEDGEMENTS
Funding:
• NIAID
Grant
#RO1AI050038
• UCOP
Pacific
Rim
Research
Program
45. Good
San
Good WQ
Bad WQ
Bad
San
Highest potential for
recontamination
High potential for
recontamination
Intermediate potential
for recontamination
Lowest potential for
recontamination
improve water
storage &
sanitation
improve water
storage
integrated
approach
improve wq +
water storage