1. Systematic Literature Reviews and Development of Distributions of Viral Densities in Raw Wastewater
Sorina E. Eftim, PhD MHS
May 18, 2016

2. Outline Systematic literature reviews Bootstrapping modeling approach
Distribution of viruses in raw sewage
Norovirus
Coliphages
Conclusions

3. DISCLAIMER
The views expressed in this presentation are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency. The work described here was funded by the U.S. EPA Office of Water, Office of Science and Technology. No official endorsement should be inferred.
The results are preliminary and should not be cited or quoted.

4. What is a Systematic Review?
Systematic – Methodical in procedure, usually based on pre-designed literature search Review – An act of carefully looking at or examining the scope, quality, and data availability
Overall Goal: To compile relevant scientific information that could potentially support future RWQC for coliphage.

5. Identifying the Literature: Process Overview and Results (1)

6. Identifying the Literature: Process Overview and Results (2)
COLIPHAGE
Literature Search 1:
406 Unique Results
Literature Search 2:
45 Unique Results
Check overlap with previous coliphage bibliographies
Sort by Water Type1
Primary Screening
(Title and Abstract):
222 Unique Results
Ambient: 117
Influent: 85
Effluent: 98
1Some studies report multiple water types
Other: 34
Data Availability:
11 studies included in project
(7 studies with somatic,
7 studies with
male-specific)
Study Quality: 43 studies pass quality
Validated Methods2
EPA methods 1601, 1602
ISO & 2
Standard Method 9224F
Full-text Review
Study scope: 81 pass scope
Quantitative data for coliphage in influent
Request Data from Authors
2All Host Strains Accepted

7. Bootstrapping Modeling Approach

8. Bootstrapping Modeling Approach (1)
Data are heterogeneous, thus a meta-analysis approach provides more comprehensive information than individual studies
Model virus concentration by estimating the mean and standard deviation (and 95% confidence intervals) of the overall distribution by bootstrapping the available data
Assumptions
Study-specific densities are representative subset of the “true” overall density
Values below LOD or above Too Numerous to Count (TNTC) are substituted with reported LOD (and/or TNTC values)

9. Bootstrapping Modeling Approach (2)
Location and season-specific distributions
Stratified sampling by location
Four regions: North America, Europe, Oceania, and Asia
Draw 10, 10, 5, & 5 samples from each location (n=30)
Stratified sampling by season
Four seasons
Draw 10 samples from each season (n=40)
Overall distribution accounting for location (or season)
Stratified sampling with unequal weights
Divide the entire data into different strata
Randomly select data proportionally from different strata

10. Bootstrapping Modeling Approach (3)
Advantages:
Method accounts for:
the amount of data reported by each study
the various geographic locations and various seasons over which the samples were taken
within and between study heterogeneity
Non-parametric approach avoids specifying and fitting a certain parametric distribution for the entire data
Limitations:
Method might not be appropriate when studies have limited number of quantifiable samples (i.e., small studies or studies with many values below limit of detection [LOD])
There are other complex approaches to address the presence of left- and right- censored data, and to deal with presence of heterogeneity

11. Norovirus distributions
Genogroups combined

12. Norovirus Data Data extracted from 15 studies; data in log10 GC/L
Locations: 12 countries (4 regions)
Europe: Norway, Sweden, Switzerland, Spain, France, Finland, Ireland
North America: USA, Canada
Asia: Japan, Singapore
Oceania: New Zealand
Type N
Mean
Median
Range
Combined
850
4.80
4.74 GI
389
4.59
4.07
GII
461
4.98
5.00
*GI (Genogroup I); GII (Genogroup II)

13. Norovirus Observed Densities by Locations and by Seasons
Europe
North
America
Asia
Spring
Summer
Fall
Winter
Oceania

14. Bootstrapping Results by Location and by Season
North America
Oceania
Europe
Asia
Summer Fall Winter Spring

15. Bootstrapping Summary Results
Analysis N
Mean (SD)
log10 GC/L
North America
219
3.9 (1.1)
Europe
549
5.1 (1.6)
Oceania 28
3.4 (1.0)
Asia 54
6.4 (0.7)
Spring (Mar to May)
208
5.3 (1.6)
Summer (Jun to Aug)
180
4.3 (1.3)
Fall (Sep to Nov)
158
4.1 (1.2)
Winter (Dec to Feb)
304
5.1 (1.5)
Accounting for location
4.6 (1.6)
Accounting for season
4.7 (1.3)

16. Norovirus distributions
Genogroup specific results

17. Norovirus Density PlotsGI
GII

18. Norovirus Observed Densities by Locations and by Seasons
GI in Pink
GII in Teal

19. Locations GI GII Seasons GI GII North America Oceania Europe Asia
Fall Summer GI
GII
Summer Fall Winter Spring
Winter Spring

20. Bootstrapping Summary Results
Analysis N
Mean (SD) GI
GII
North America
112 | 107
3.3 (0.7)
4.7 (1.0)
Europe
244 | 305
5.2 (1.6)
5.0 (1.6)
Oceania
15 | 13
3.1 (0.8)
3.8 (1.2)
Asia
18 | 36
6.4 (0.4)
6.4 (0.8)
Spring (Mar to May)
96 | 112
5.2 (1.6)
5.4 (1.6)
Summer (Jun to Aug)
81 | 99
3.9 (1.2)
4.6 (1.4)
Fall (Sep to Nov)
80 | 78
4.0 (1.3)
4.2 (1.3)
Winter (Dec to Feb)
132 | 172
4.9 (1.8)
5.3 (1.4)
Accounting for location
4.4 (1.7)
4.9 (1.4)
Accounting for season
4.5 (1.6)
4.9 (1.5)

21. Coliphage distributions
Male-specific (F-specific) and Somatic

22. Coliphage Data
Data extracted from 11 studies; data in log10 counts/L (PFU/L and MPN/L)
Locations: 8 countries (4 continents)
Europe: Netherlands, Italy, Spain
North America: USA, Canada
Asia: Japan, China
Africa: South Africa
Type N
Mean
Median
Range
Combined
439
6.29
6.43
F-specific
243
6.05
6.23
Somatic
196
6.58
6.95
Male-specific
Somatic

23. Coliphage Observed Densities by Locations and by Seasons
Male-specific in Pink
Somatic in Teal

24. Bootstrapping Results
Location
Somatic
Male-specific
Asia
North America
Europe
Asia
Europe
North America
Africa

25. Bootstrapping Results
Season
Male-specific
Somatic
Winter
Fall
Spring
Spring
Fall
Winter
Summer
Winter
Summer
Season
Spring
Spring

26. Bootstrapping Summary Results
Analysis N
Mean (SD)
Male-specific
Somatic
North America
213| 43
6.0 (1.0)
5.2 (1.4)
Europe
9 | 81
6.6 (0.3)
6.8 (0.7)
Africa
N/A | 8
N/A
6.1 (0.7)
Asia
21 | 64
6.1 (0.3)
7.4 (0.3)
Spring (Mar to May)
90 | 57
5.6 (1.1)
6.7 (1.4)
Summer (Jun to Aug)
28 | 72
5.6 (1.3)
6.2 (0.7)
Fall (Sep to Nov)
37 | 28
6.3 (0.5)
6.6 (0.7)
Winter (Dec to Feb)
88 | 39
6.4 (0.5)
6.9 (0.3)
Accounting for location
5.6 (1.1)
6.7 (1.4)
Accounting for season
5.6 (1.3)
6.2 (0.7)

27. Conclusions – Bootstrapping
Heterogeneity in study-specific curves, location, and season suggests that one overall parametric distribution function might not represent data well.
A meta-analysis approach provides more comprehensive information than individual studies.
Bootstrapping provides a complete representation of data with no loss of information.

28. Conclusions – Viral distributions
Norovirus in raw wastewater are well represented by distributions with ranges in the order of 104 – 106 GC/L
Somatic coliphages in raw wastewater are well represented by distributions with ranges in the order of 105 – 107 counts/L*
Male-specific coliphages in in raw wastewater are well represented by distributions with ranges in the order of 105 – 106 counts/L*
*Data are expressed in log10 Counts/L (PFU/L and MPN/L)

29. Contributors Sharon Nappier, PhD MHS, EPA/OW/HECD
Tao Hong, PhD, ICF International
Audrey Ichida, PhD, ICF International
Isaac Warren, ICF International
Jeffrey A. Soller, Soller Environmental

30. Thank you!
Sorina Eftim, PhD MHS