1. Risk assessment and health based targets for microbial water quality Prof. Nicholas J. Ashbolt School of Civil & Environmental Engineering University of New South Wales Sydney, Australia Cooperative Research Centre for Water Quality and Treatment

2. Public health based water quality guidelines  Health impacts of waterborne disease vary in severity and impact  (Marginal) cost-effectiveness of additional measures decreases steadily  Decide which risk is tolerable… against current background of costs and population health  Then define tolerable exposure

3. WHO harmonised approach for provision of safe water HEALTH TARGETS Basic control approaches Water quality objectives Other management objectives Define key risk points and audit procedures for overall system effectiveness Define analytical verifications (process, public health) Define measures and interventions (requirements, specifications) based upon objectives Risk Assessment Assess environmental exposure PUBLIC HEALTH OUTCOME RISK MANAGEMENT (HACCP) Tolerable risk Fewtrell & Bartram 2001

4. Pathogens are primarily from faeces, with various modes of transmission: Excreta from humans and animals Human Shellfish Crops Aerosols Oceans and Estuaries Oceans and Estuaries Rivers and Lakes Rivers and Lakes Irrigation Solid Waste Landfills Solid Waste Landfills Sewage Land Runoff Land Runoff Recreation Water Supply Water Supply Groundwater Adapted from Charles P. Gerba et al. 1975

6. Cholera: cases & case fatality rates El Tor biotype of Vibrio cholerae Cholera outbreaks appear to originate in travel and trade from Bangladesh to Pakistan and then spread throughout south-central Asia. Classic O1

7. E. coli outbreak in Walkerton, Canada

8. Agents associated with waterborne outbreaks (CDC, USA) AGI - acute gastrointestinal illness

9. Epidemic to endemic illnesses as detected by epidemiologic studies Number of Cases Time Threshold for detection for an outbreak Endemic rate Outbreak detected Undetected outbreak Hyperendemic Sporadic Frost et al. (1996) J. Am. Wat. Wk. Assoc. 88(9):66-75

10. How much diarrhoea is tolerated: developed regions?  Current rates (by strict definition): USA 1.05 episodes per person a year (Mead et al. 1999) UK 0.19-0.55 episodes per person a year (Wheeler et al. 1999) Australia 0.92 episodes of gastroenteritis per person a year (Hellard et al. 2001)  Gastro in USA, hospitalisation rate of 4.4%, fatality rate 0.03% for community cases Foodborne illness was estimated to cause 36% of cases, 34% of hospitalisations and 67% of deaths (Mead et al. 1999)

11. So how much diarrhoea is tolerable?  To translate the risk of developing a specific illness to disease burden per case, the metric DALYs is used  Disability Adjusted Life Years (DALYs) mortality: years of life lost (YLL) morbidity: years lived with disability (YLD) DALY = YLL + YLD  Conceptually simple: health loss = N x D x S  N = number of affected persons  D = duration of adverse health effect  S = measure for severity of the effect

12. YLD Hypothetical example of DALY Residual disability Premature death Acute (infectious) disease YLL

13. Disease burdens for different water contaminants

14. Reference Pathogens  Viruses Rotavirus (Adenovirus)  Bacteria Campylobacter & STEC (developed) STEC and Vibrio cholerae (developing)  Parasitic Protozoa Cryptosporidium/Giardia Entamoeba histolytica  Helminths Ascaris lumbricoides

15. Using DALYs to derive Water Quality Targets (WQT)  Tolerable mortality risk 10 -5 per lifetime  1.4 x 10 -7 per year (for a lifespan 70 years)  Corresponds to  10 -6 DALY per year.person  Disease burden for e.g. rotavirus in developed countries is 0.014 DALY per case  Tolerable incidence of rotavirus associated gastroenteritis is 7.14 cases per Mpy  Use consumption data and Dose-Response model to derive WQT, accounting for sensitive fraction in the population

16. Linking tolerable disease burden and source water quality: Ref. pathogens Parameter Cryptospor- idium Campylo- bacter Rotavirus a Pathogens/L in source1010010 Health outcome target ( per person a year) 10 -6 DALYs Annual risk of diarrhoeal illness b 1 per 16001 per 40001 per 11 000 Drinking-water quality1 per 1600 litres 1 per 8000 litres 1 per 32 000 litres Performance target c 4.2 log 10 units 5.9 log 10 units 5.5 log 10 units a Data from high-income regions. In low-income regions, severity is typically higher, but drinking-water transmission is unlikely to dominate. b For the susceptible population (100%, 100%, 6% respectively). c Performance target is a measure of log reduction of pathogens based on source water quality (e.g. 4.2 log 10 = 99.994% reduction).

17. Linking tolerable disease burden and source water quality: Ref. pathogens Parameter Cryptospor- idium Campylo- bacter Rotavirus a Pathogens/L in source1010010 Health outcome target ( per person a year) 10 -6 DALYs Annual risk of diarrhoeal illness b 1 per 16001 per 40001 per 11 000 Drinking-water quality1 per 1600 litres 1 per 8000 litres 1 per 32 000 litres Performance target c 4.2 log 10 units 5.9 log 10 units 5.5 log 10 units a Data from high-income regions. In low-income regions, severity is typically higher, but drinking-water transmission is unlikely to dominate. b For the susceptible population (100%, 100%, 6% respectively). c Performance target is a measure of log reduction of pathogens based on source water quality (e.g. 4.2 log 10 = 99.994% reduction).

18. Linking tolerable disease burden and source water quality: Ref. pathogens Parameter Cryptospor- idium Campylo- bacter Rotavirus a Pathogens/L in source1010010 Health outcome target ( per person a year) 10 -6 DALYs Annual risk of diarrhoeal illness b 1 per 16001 per 40001 per 11 000 Drinking-water quality1 per 1600 litres 1 per 8000 litres 1 per 32 000 litres Performance target c 4.2 log 10 units 5.9 log 10 units 5.5 log 10 units a Data from high-income regions. In low-income regions, severity is typically higher, but drinking-water transmission is unlikely to dominate. b For the susceptible population (100%, 100%, 6% respectively). c Performance target is a measure of log reduction of pathogens based on source water quality (e.g. 4.2 log 10 = 99.994% reduction).

19. Linking tolerable disease burden and source water quality: Ref. pathogens Parameter Cryptospor- idium Campylo- bacter Rotavirus a Pathogens/L in source1010010 Health outcome target ( per person a year) 10 -6 DALYs Annual risk of diarrhoeal illness b 1 per 16001 per 40001 per 11 000 Drinking-water quality1 per 1600 litres 1 per 8000 litres 1 per 32 000 litres Performance target c 4.2 log 10 units 5.9 log 10 units 5.5 log 10 units a Data from high-income regions. In low-income regions, severity is typically higher, but drinking-water transmission is unlikely to dominate. b For the susceptible population (100%, 100%, 6% respectively). c Performance target is a measure of log reduction of pathogens based on source water quality (e.g. 4.2 log 10 = 99.994% reduction).

20. Issues in adapting risk-based performance target setting to local circumstances  The choice of reference pathogens is mainly based on availability of data  The pathogens illustrated may not be priority pathogens in all regions of the world  Wherever possible, country- or site-specific information should be used in assessments of this type  If no specific data are available, an approximate risk estimate can be based on default values in guidelines  Presented point estimates do not account for variability and uncertainty

21. Accounting for poor performance in the risk assessment  Nominal performance can be described stochastically (i.e. as a distribution)  Similarly failure duration and effect can be described stochastically Distribution 31.8% Groundwater 18.2% Unknown 9.1% Surface 4.5% Treatment deficiency 35.4% Failures leading to Waterborne illness in the USA

22. Example failure in raw water reservoir during a rain event

23. Validation of reservoir models www.cwr.uwa.edu.au/~ttfadmin/model/inflow Myponga (26 GL, 30m) Burragorang (2000GL, 90m ) measuredpredictedmeasuredpredicted Inflow °C1010.5 Insertion °C 14.714.9 12.5-12.7 12.6 Travel time 30 h27 h7 d6.5 d Inflow dilution 7.1 x8 x1.9-2.9 x2.5 x

24. Example of a failing treatment 1 1 Teunis et al. (2004) Short-term fluctuations, WHO, Geneva a)Schematic of failing process, say 10% of operational time b)Performance for pathogen removal, bimodal outcome 100 pathogens entering barrier 1 log nominal 0.3 log failure

25. Daily risks (grey curves), Annual risks: with and without failure

26. In summary HEALTH TARGETS Basic control approaches Water quality objectives Other management objectives Define key risk points and audit procedures for overall system effectiveness Define analytical verifications (process, public health) Define measures and interventions (requirements, specifications) based upon objectives Risk Assessment Assess environmental exposure PUBLIC HEALTH OUTCOME RISK MANAGEMENT (HACCP) Tolerable risk

27. Rationale for multiple barriers: Probability of reaching at least 2 log 10 removal (P<0.01) Each barrier (p 0 =0.1, p 1 =0.03) and probability of failure P f = 0.1, 0.3, 0.5, 0.7, and 0.9. Note that at least two but no more than seven barriers necessary.

28. Combining WQT (  5 µDALY), pathogen # in source waters & system performance  Estimate the percent removal necessary for safe water provision Need for catchment improvement, and/or treatment  QMRA also provides data on Variability (inherent to the system) Uncertainty (lack of knowledge of system) Target ranges at critical control points Selection of appropriate new systems Are health targets met

29. Links between WSP & QMRA

30. The principles of HACCP (hazard analysis critical control point) Construct & valid process flow diagram 1. Identify hazards and preventative measures 2. Identify Critical Control Points 3. Establish critical limits 4. Identify monitoring procedures 5. Establish corrective action procedures 6. Validate/verify HACCP plan 7. Establish documentation and record keeping The seven principles of HACCP Initial steps Not drinking water, but upstream for key index organisms

31.  In developing regions ~ 2/y gastro rate and for expatriates from developed regions ~ 3/y  An appropriate health target from waterborne exposure could be 0.8% gastro, given that is what is currently tolerated today in developed regions  With an additional safety factor ~ 0.01% (10 -2 per annum illness risk) would be about as strict as could be defensible.  Particularly given that extra community dollars spent on improved water treatment may take dollars away from other (more beneficial) areas of health protection.

32. What is tolerable risk for other industries? How does it compare to 1 infection/100 people a year?

33. Target risk based on consequences of failure for industries K, A & W are factors regarding the redundancy of the structure T is the life of the structure; n is the number of people exposed to risk Consequence of Failure (lives lost) 1101001000 Annual Probability of Failure 10 10 -1 10 -2 10 -3 10 -4 10 -5 10 -6 Merchant shipping Mobile drill rigs Drowning - associated with large boats Dams Commercial aviation Construction Ind. Res. & Info. Assoc. Equation Allen’s Equ. *

34. Linking tolerable disease burden and source water quality  1 µDALY per year ParameterUnitsCryptos- poridium Campy- lobacter Rota- virus Raw water quality (C R ) Organisms/L1010010 Treatment effect needed to reach tolerable risk (PT) Percent reduction 99.99499.9998799.99968 Drinking-water quality (C D ) Organisms/L 6.3 × 10 -4 1.3 × 10 -4 3.2 × 10 -5 Consumption of unheated drinking- water (V) L/d111 Exposure by drinking water (E) Organisms/d6.3 × 10 -4 1.3 × 10 -4 3.2 × 10 -5 C D = C R × (1 -PT) E = C D × V