I'Douglas, P'Eng' City of Ottawa

1
I.Douglas, P.Eng. City of Ottawa
Quantifying Microbiological Risks in Drinking
Water
Canadian Water Network Victoria, B.C. June
25th, 2008
2
Acknowledgements

• Stephanie McFadyen Health Canada, Microbiology
• Water, Air, Climate Change Bureau
• Emma Hartnett Decisionalysis Risk Consultants

3
a step back in time

• Do ye hear the children weeping, O my
  brothers?
• For what? for a glass of cold water, fresh from
  the tap. But that would make them ill, poor
  creatures, give them chills and typhoid. Only
  today a woman said to me, My little son had the
  fever for several weeks. He is such a boy for
  drinking water, and, it has given him the fever
• signed, ONE WHO CAN VOTE
• Letter to editor, Ottawa Journal
  December 24th, 1888

P.S. Ottawa 1911-1912 waterborne outbreak of
typhoid gt140 deaths
4
How can we quantify microbial risks in drinking
water?
microbial pathogens




health risk


tolerable level of risk ?

treatment
5
microbiological safety of drinking water

• detection of pathogens in treated drinking water?
  nearly impossible, reactive at best
• historical use of indicator organisms in treated
  water ? spatial temporal challenges! eg.
  99.9999998
• outbreaks of waterborne illness, even in the
  absence of adverse indicators
• move towards preventive design/barrier approach,
  rather than reactive
• apply risk concepts from food pharmaceutical
  industry (HACCP, QMRA)

risk-based management
cant test every drop ! cant do product recall !
6
Water treatment applied to reduce pathogen risk
to safe or acceptable levels
acceptable risk level (WHO)
level of detection reportable GI
raw waters
treated waters
high risk
low risk
10-12
10-9
10-6
10-3
100
103
106
Health Risk
7
What is QMRA?
QMRA Quantitative Microbial Risk Assessment

• Method to evaluate the risks of microbial
  pathogens in a drinking water supply, and
  estimate the impact on population health

How much illness can be attributed to drinking
water? What is the effect of treatment?
treatment
lab analysis
microbiology
statistics
population health
engineering
log removals
CT disinfection
epidemiology
8
WARNING you are now entering the world of
uncertainty probability
RISK HAPPENS !

• Prevention of death is quite difficult. In
  fact, the probability of dying is approaching
  100
• ( Dr.L.Bonneux, Netherlands )

9
What about uncertainty? fuzzy cloud
pathogen recovery?
infectious?
sample right time?
risk
log-removal?
optimal?
source
particle clumping?
treated
water consumption?
virulence of strain?
health status?
juice, coffee, tea?
ingested
10
Key steps in QMRA
Conc. of pathogens in raw water
Calc. overall treatment log-removal
Pathogen conc. in treated water
Pathogens ingested per day
P (infection) given dose
Annual illnesses in population
Annual health impact
P(illness) given infection
11
Choose a number of index or reference pathogens
choose (5) reference pathogens
Raw Water
Pseudomonas
E.coli O157
Cryptosporidium

• source water data
• present in most surface impacted ground waters
• vary in size, Cl2 resistance, treatment removal
• dose-response data

Vibrio
HepA
E.coli
Giardia
Poliovirus
Mycobacterium
Rotavirus
Campylobacter
Legionella
Adenovirus
Yersinia
Shigella
Salmonella
NRC images
12
Health Canada QMRA Model

• Consider risks from (5) pathogens
• Cryptosporidium parvum
• Giardia lamblia,
• Rotavirus,
• pathogenic E.coli,
• E.coli O157H7
• (4) physical removal methods\
• conventional (coag/sed/filtration)
• slow-sand filtration
• microfiltration
• ultrafiltration
• (5) primary disinfection methods
• chlorine
• chloramine
• ozone
• chlorine dioxide
• ultra-violet (UV)
• Literature references used for specific pathogen
  removals inactivation

13
Health Canadas QMRA Model User Input Screen
population
Crypto Giardia Rotavirus E.coli default tables
treatment conditions (CT, pH, T, Cl2 residual)
14
part 1 probability that a microbial pathogen
will pass from source water to tap
1 .0 L/day sample
pathogens ingested
removal through treatment
pathogen conc. In raw water
15
part 2 probability of infection, given a dose of
pathogens

• Exponential

Crypto
Beta-Poisson
Rotavirus
16
Health impacts in the population?( cases of
illness, burden of disease)
Crypto
?
E.coli
Microbial risks usually expressed in number of
illnesses in population
?
17
Estimating Burden of Disease DALYs

• but, different pathogens can have a range of
  health outcomes with very different implications
• ? severity (0-1) x duration (time)
• World Health Organization has developed the DALY
  metric to express health outcomes in terms of the
  human impact DALY Disability Adjusted Life Year

A few comparisons DALYs per case Crypto
1.7 x 10-3 Rotavirus
8.5 x 10-3 E.Coli ENT 25 x
10-3 Colon cancer 3.4 Liver cancer
11.0
population health target of 10-6 DALYs per year
per person (drinking water)
18
How does the 10-6 DALY risk level compare?

• 10-6 DALYs per year per person is equivalent to
• GI Illness
• 5.9 x 10-4 annual risk of Cryptosporidium
  illness
• 1.2 x 10-4 annual risk of Rotavirus illness
• 0.4 x 10-4 annual risk of E.coli O157 illness
• Cancer
• 2.0 x 10-5 lifetime (70 yr) risk of colon
  cancer
• 6.4 x 10-6 lifetime (70 yr) risk of liver
  cancer

19
QMRA Case Study evaluate effects of water
treatment on waterborne illness in City of Ottawa

• Microbial risk level for current treatment
• Which pathogen is driver of risk?
• Impact of individual treatment barriers
  (failure?)
• Health impact of future treatment changes

20
Q1 Which pathogen is driving the health risk
in Ottawa?
21
Q2 What is the relative importance of each
treatment barrier?
10-6 DALYs/pp/yr
22
Q3 What level of pathogens can we handle in the
raw water?
Maximum
Average
23
Estimated annual illnesses for normal treatment
and failure events

• annual risk of illness 7.7 x 10-8
• DALYs per year (individual) 1.3 x 10-10
• WHO health target (individual) 1.0 x 10-6

4 orders of safety

• partial FF (14) filters 5.5-log (4) at 2.6-log
• full FF (18) filters 1.6-log

24
Reportable GI illnesses in City of Ottawa all
causes(2000 2006)

• Number of annual cases
• Crypto 15
• Giardia 143
• E.Coli Verotox 22
• Amebiasis 74
• Campylobacter 256
• Salmonella 147
• Shigella 52
• Yersinia 15
• Total GI 725 confirmed
• cases per year

Ottawa population 725 cases/yr of acute GI
reported x 313 actual for each (1) reported
case 227,000 cases/yr or if we use Canada
estimate of 1.3 cases pp/yr of acute GI
illness 1,040,000 cases/yr Contribution from
drinking water 0.06 cases/yr
25
Q4 Compare the risk of illness during normal
treatment and failure events
Estimated annual illnesses for various treatment
conditions

• partial FF (14) filters 5.5-log (4) at 2.6-log
• full FF (18) filters 1.6-log

26
Q5 How does the microbial risk vary as we apply
treatment (eg. chlorine disinfection) ?
Current
O.Reg.
27
Advantages of Using QMRA

• Science health-based approach literature
  values
• Case specific uses actual pathogen levels and
  treatment conditions to calculate risk
• Flexible tolerable risk level can be adjusted
  for a given population
• Transparent risk assessment and assumptions are
  clearly laid out
• Wide application can be used to provide insight
  for many risk-based decisions (source, treatment,
  public health, inspection, etc.)

28
Challenges with QMRA

• Source water pathogen data for best results,
  require data for Crypto, Giardia, virus
  concentrations
• Distribution system risks not yet defined
• Uncertainty order of magnitude results
  require care in interpreting health impacts

29
Next Steps

• Model validation using data from waterborne
  outbreaks
• near misses
• Web-based version of Health Canadas QMRA model
• Compare microbial with chemical risks using
  DALYs to optimize treatment for minimizing
  overall population health risk

DBPs (cancer)
pathogens (GI illness)
health risk (DALYs)
disinfection
30
Thank you
Ian Douglas City of Ottawa ian.douglas_at_ottawa.ca
31
Probabilistic approach vs. single-point
calculation (eg. Giardia)
90th percentile
raw water Giardia conc.
treatment log-removal
pathogens ingested
probability of infection
1 in 10,000 risk of infection
SWTR
32
What is health?

• Individuals capability to function
• well physically, mentally, and socially
• (Havelaar Melse, World Health Organization)

33
What is an acceptable level of health risk?
level of acceptable risk is a value judgement,
depends on culture, societal values, and other
risk factors such as economy, security,
war, health care, hunger, etc.
WHO has set 1x10-6 DALY/yr/pp as an acceptable
health target for drinking water
1 in 10,000 risk of GI illness 10-5 to 10-6
lifetime risk of cancer
34
Applications of QMRA

• Treatment Engineering - assessment of treatment
  barriers
• Operations impact of normal vs. upsets, trigger
  levels
• Public health population impacts, boil
  advisory, health policy
• Capital planning cost/benefit for new or
  additional treatment
• Regulatory guidelines, micro. vs. chemical,
  treatment requirements
• Inspection evaluation of system risks,
  prioritization
• Source Water Protection impact of reducing
  pathogen levels

35
What is the relative importance of each
treatment barrier?
10-4 Illness risk
10-6 DALYs/pp/yr
36
WHO health loss measured in DALYs (disability
adjusted life years) severity (0 ? 1.0) and
duration
asthma
(Havelaar Melse, 2003)
37
Examples of severity weights for selected health
outcomes ( 0 1 ) scale
The Global Burden of Disease - Murray C.J.L,
Lopez A.D.(World Health Organization, 1996)
A few comparisons DALYs per 1,000
cases Crypto 1.7 x
10-3 Rotavirus 8.5 x 10-3 E.Coli
ENT 25 x 10-3 Colon cancer
3.4 Liver cancer 11.0
38
Why are we here?
The primary purpose of drinking water
treatment is the removal of waterborne pathogens
and the management of risks that arise from
their removal
39
Internationally - QMRA in Drinking Water

• Dutch Drinking Water Decree (2001) Owners of
  water supply must demonstrate less than 1 per
  10,000 risk of infection per year from
  Cryptosporidium, Giardia, and Enterovirus
• Australian Guidelines for Drinking Water Quality
• WHO Guidelines for Drinking Water Quality (2004)
• European Unions Microrisk project (2006)
• Health Canada develop QMRA model (2006) QMRA
  used in new guidelines for Protozoa Viruses
• Others (UofT, Guelph, Montreal)

40
QMRA Model User Input Screens
population
source water pathogens (5) default table (eg.
pristine, heavily impacted, etc.)

treatment conditions (CT, temp, pH, dose,
logremovals)
41
How does QMRA work?
42
todays presentation
how does QMRA work?
drinking water contribution?
what is a DALY?

• what is health?

how we can apply QMRA?
how much diarrhoea in Canada?
43
Public health surveillance tip of the iceberg
GI cases reported vs. actual (2005 Public
Health Agency of Canada survey in Ontario B.C.)

• 1 positive case
• reported
• 1.24 positives reported locally
• 1.56 stools positive
• 14 stools tested
• 15 submit the stool
• 19 have stool samples requested
• 73 cases visit physician
• 313 community cases of GI illness

in Canada, estimated 1.3 cases of acute GI
illness/year/person
C.Majowicz et al, 2005 Can.J.Public Health
96178-81
44
What is the health benefit to communityof water
treatment?
In Canada, estimated cost of 1,089 per case
acute GI illness Majowicz et al, 2006
2.0 million cases of acute GI illness/yr
Cost
0.06 cases of acute GI illness/yr
45
How does pathogen risk change with seasons?
46
Estimated annual illnesses for normal treatment
and failure events

• partial FF (14) filters 5.5-log (4) at 2.6-log
• full FF (18) filters 1.6-log

47
What is the impact of a treatment failure? (eg.
loss of Cl2 disinfection for 12 hoursMar.11th)
7 cases (E.coli-ENT)
Illnesses
what is an appropriate trigger for plant
operations?
48
remember ?
microbial pathogens




health risk


tolerable level of risk ?

treatment
49
What is the impact of adding UV as an additional
treatment barrier?
current CT
DALYs/yr
50
What is the impact of adding UV as an additional
treatment barrier?
current CT
DALYs/yr
with UV
51
Next Steps

• Model validation using data from waterborne
  outbreaks
• near misses
• Web-based version of Health Canadas QMRA model
• Compare microbial with chemical risks using
  DALYs to optimize treatment for minimizing
  overall population health risk

52
one final thought

• drinking water is
• essential for health
• revitalizing, cleansing
• sparkling clear refreshing
• sustains us
• feel good about it !
• drinking water quality of life

53
What is the probability of ingesting x
pathogens,given a mean tap water concentration ?
Poisson PDF
0.1 per Litre
1 per Litre
10 per Litre
54
Ottawas water supply
N
Britannia W.P.P. 360 ML/d
Ottawa River - pristine, wilderness
- soft, high colour/TOC - microbial pathogens
Lemieux Island W.P.P. 290 ML/d
55
Example Cryptosporidium

• (2) typical outcomes diarrhoea or death
• calculate severity duration for each outcome

A few comparisons (DALYs per case) Crypto
0.0017 Rotavirus 0.0085 E.Coli
ENT 0.025 Colon cancer 3.4 Liver
cancer 11.0
56
(5) microbial reference pathogens

• present in most surface GUDI waters
• vary in size, Cl2 resistance, treatment removal
• dose-response data available

NRC images
Photo credits protozoa, H.D.A Lindquist, U.S.
EPA Rotavirus, F.P. Williams, U.S. EPA
57
Dose-Response for Pathogens in QMRA Model
Rotavirus
E.coli ENT
Giardia
Crypto
E.coli NON
58
probabilistic nature of model (eg. Crypto)
Raw water risk
treated water risk
59
Overview of model
60
Ottawa typhoid epidemic of 1911/1912
Typhoid death rate in City of Ottawa (1905 1917)
waterborne outbreak (141 deaths)
chlorine
chloramine
61
a long time agoin a galaxy (not) so far away!

• Water, water everywhere nor any drop to drink.
• Skilful analyst and microscopists have taken
  water obtained from our tapsconclude that the
  presence of an unusual number of vegetable and
  animal organisms render it unfit to be used as a
  beverage. In must be boiled, and well boiled, to
  render it safe to be drunk.
• Do ye hear the children weeping, O my brothers
  
• For what? for a glass of cold water, fresh from
  the tap. But that would make them ill, poor
  creatures, give them chills and typhoid. Only
  today a woman said to me, My little son had the
  fever for several weeks. He is such a boy for
  drinking water, and, it has given him the fever
• signed, ONE WHO CAN VOTE
• Ottawa Journal December 24th, 1888

62
How does each pathogen respond to disinfection
treatment ?
63
Applying QMRA case studyPopulation
800,0001.0 L/day daily water consumptiondata
for normal plant operation
cfu/100mL
1Primary Disinfection t45 min Cl20.62 mg/L
pH6.0, Temp10C
64
Probability of illness, Pill

• probability of illness (symptoms), given
  infection
• P(ill) P(inf/dose) x P(ill/inf)
• determine of illnesses in population

values from health literature and analysis of
past waterborne outbreaks
65
Q1 Is the current treatment process adequate?
Total per year
illnesses per year (population)
0.06 annual risk of illness (individual)
7.7 x 10-8 DALYs per year (population) 1.0
x 10-4 DALYs per year (individual) 1.3 x
10-10 WHO health target (individual) 1.0 x
10-6
4 orders of safety
66
(No Transcript)
67
Model Screens
literature values used, equations, references
68
Probabilistic model key steps
Lognormal dist.

• Pathogen conc. (mean,std.dev)
• in raw water
• Treatment log (define barriers
• Reduction operating
• conditions)
• Pathogen conc. (mean conc.)
• In treated water
• pathogens (Poisson dist.)
• ingested(daily)
• Pinfection, daily (dose-response)

69
Estimating Burden of Disease What is a DALY?

• DALY Disability Adjusted Life Years
• Expresses health impact in terms of both quality
  quantity of life
• DALY YLD LYL
• aggregate measure that takes into account the
  duration and severity of the illness
• 1 DALY is equivalent to 1 year of healthy life
  lost

A few comparisons DALYs per 1,000
cases Crypto 1.7 Rotavirus
8.5 E.Coli ENT 25.0 Colon cancer
3,400 Liver cancer 11,000
70
Public health surveillance for reportable
diseaseseg. Giardia
71
Public health data for reportable
diseasesE.coli-ENT
72
What is an acceptable level of health risk for
drinking water?

• 10-4 annual risk (1 in 10,000) of GI illness
• (US-EPA, Netherlands)
• 10-5 to 10-6 lifetime cancer risk
• (Health Canada, US-EPA, WHO)
• 10-6 DALYs per year per person
• (World Health Organization)

73
What is a negligible level of illnessin the
population? Can we achieve risk level below
that detected?

• EXAMPLE
• Cryptosporidium in New York City (10 million
  pop.)
• 10-4 risk level 1 in 10,000 infections per year
• people infected 1,000
• symptomatic cases/yr 600
• cases that consult doctor 200
• cases with stool analysis 50
• cases w. adequate test selected 5
• test perform. data handling 3
• PHAC level of GI illness reporting, etc.

74
City of Ottawa detectable illness
ratespopulation 750,000

• Crypto 10-6 DALY pp/yr risk level 5.9x10-4
  risk of illness pp/yr
• 441 cases/yr in population
• But, 24 cases normally reported/yr (2 per month)
• 7,512 cases/yr actual (PHAC figure of 3131)
• If assume 50 increase in case rate to barely
  detectable
• 36 cases/yr reported (3 per month)
• 11,268 cases/yr (PHAC)
• Difference in cases (if allocated to drinking
  water) 11268 7512 3756
• additional actual cases/yr caused from drinking
  water
• 3,756 / 750,000 5.0 x10-3 risk of illness
  multiply by DALY/case (1.7x10-3)
• 8.5 x 10-6 DALY/pp/yr level 10x higher than
  WHO target to be barely detectable

75
City of Ottawa detectable illness
ratespopulation 750,000

• Giardia 10-6 DALY pp/yr risk level 5.9x10-4
  risk of illness pp/yr
• 441 cases/yr in population
• But, 216 cases normally reported/yr (18 per
  month)
• 67,608 cases/yr actual (PHAC figure of 3131)
• If assume 50 increase in case rate to barely
  detectable
• 324 cases/yr reported (27 per month)
• 101,412 cases/yr (PHAC)
• Difference in cases (if allocated to drinking
  water) 33,804 additional cases/yr (caused by
  drinking water)
• 33,804/750,000 0.045 risk of illness
  multiply by DALY/case (1.7x10-3)
• 7.6 x 10-5 DALY/pp/yr level 100x higher
  than WHO target to be barely detectable

76
Risks

• of deaths in Canada by cause (1994)
• cardiovascular diseases 37.9
• cancer 28.2
• respiratory diseases 8.9
• external causes 6.4
• digestive diseases 3.7
• endocrine others 3.2
• nervous system 2.9
• genito-urinary diseases 1.6
• infectious diseases 1.5
• other diseases 5.7

100.0
77
Risks

• Shark attacks vs. lightning
• Average per year
• (from 1959 1994 stats reported in North
  Americacoastal states only to make comparison)
• lightning strikes 179.7
• lightning fatalities 44.9
• shark attacks 11.4
• shark attack fatalities 0.5
• Mode of travel chart (auto, motorcycle, bus,
  etc.)
• Mountain climbing, parachuting, boxing, scuba,
  snowmobiling, hiking, skiing racing, etc.

78
What is health?

• Individuals capability to function
• well physically, mentally, and socially
• (Havelaar Melse, World Health Organization)
• ability to perform activities of everyday life
• procreation, occupation, education, and
  recreation
• (Global Burden of Disease Project, WHO 1996)

79
reduction of Bacillus coli with chloramine
treatment(J.Race Ottawa, 1916)
Historical use of water treatment to reduce
pathogens to a threshold level reliance on
indicator organisms
3.0-log inactivation
80
(No Transcript)