Health Stream Literature Summary- Issue 50 - June 2008

Effect of pathogen concentrations on removal of Cryptosporidium and Giardia by conventional drinking water treatment.
Assavasilavasukul, P., Lau, B.L.T., Harrington, G.W., Hoffman, R.M. and Borchardt, M.A. (2008) Water Research, 42(10-11); 2678-2690.

Previous studies have evaluated the removal of Cryptosporidium oocysts, Giardia cysts and other emerging waterborne pathogens during conventional drinking water treatment. However the spike concentrations used in these experiments were 6-8 orders of magnitude higher than those generally observed in natural waters. There is considerable uncertainty about pathogen removals that would be achieved under more realistic and much lower spiked pathogen concentrations as particle concentration may have significant impacts on particle separation kinetics. This study was conducted to compare the removal of Cryptosporidium oocysts and Giardia cysts that were spiked at high initial pathogen concentrations and at more realistic concentrations and to describe removal as a function of spiked pathogen concentration. Also this study examined dependence of pathogen removal on other variables such a raw water quality (turbidity and temperature), alum dose and sampling method.

There were six experiments conducted from November 2004 to March 2007. The pilot plant received raw water from Lake Mendota in Madison, Wisconsin during all the experiments. Cryptosporidium and Giardia were not detected in the background raw water collected during any of the experiments. The pilot plant included two parallel treatment trains, one with a higher pathogen spike than the other. On each treatment train the raw water went through coagulation, flocculation and sedimentation. The raw water for each treatment train was spiked with a mixture of Cryptosporidium parvum oocyst and Giardia lamblia cysts. Initial pathogen concentrations ranged from 101 to 106 pathogens/L. There were a total of 9 locations sampled for each experiment. Two sampling methods were used (grab and continuous) for quantification of pathogens in the filter effluent.

To examine the effect of initial pathogen concentrations on log removal through conventional treatment, three grab sample results from each treatment train in the January 2005 experiment were used. Average Cryptosporidium oocyst log removals of 3.2 +/- 0.5 logs and 1.1 +/- 0.3 logs were achieved in the treatment trains with initial Cryptosporidium oocyst concentrations of 7.0 x 104 and 5.3 x 102 oocysts/L respectively. Average Giardia cyst log removals of 3.8 +/- 0.1 and 1.5 +/- 0.4 were achieved in treatment trains with initial Giardia cysts concentrations of 6.1 x 104 and 3.9 x 102 cysts/L respectively. At lower spike doses removal was about two orders of magnitude lower than at higher spike doses. The mean log removals in the train with the higher pathogen concentrations were significantly different at a 95% confidence levels from the mean log removals in the train with the lower pathogen concentrations (p-value = 0.03 and 0.01 for Cryptosporidium oocyst and Giardia cysts, respectively). The continuous sampling log removal results for 30 h also showed lower pathogen removals at lower spike doses. For Cryptosporidium oocysts, conventional treatment was able to remove 2.6 logs and 2.2 logs when the respective spike doses were 7.0 x 104 and 5.3 x 102 oocysts/L. For Giardia cysts, conventional treatment was able to remove 4.3 logs and 3.1 logs when the respective spike doses were 6.1 x 104 and 3.9 x 102 cysts/L.

The initial oocyst concentration could explain 44% of the variance in the log removal data for Cryptosporidium. Multiple regression analysis was performed to assess whether any other explanatory variables were significant. For Cryptosporidium, log removal was associated with raw water turbidity (slope =0.5 and p-value less than 0.01) in addition to the initial oocyst concentration. Raw water turbidity and initial oocyst concentration explained 68% of the variance in log removal values, with higher log removal seen for higher raw water turbidity. For Giardia, multiple regression analysis showed that log cyst removal was a function of initial cyst concentration, raw water turbidity and sampling method. Raw water turbidity, sampling method and initial cyst concentration explained 73% of the variation in the log removal data whereas the initial cyst concentration explained 55% of the variation.

When grab sampling and continuous sampling were compared for determining Cryptosporidium removal, the mean log removal estimated for continuous sampling was 0.3-log higher than that calculated from grab sampling however the difference between grab and continuous sampling was not statistically significant with a p-value of 0.12. The mean log Giardia cyst removal calculated from continuous sampling was 0.6-log higher than that calculated from grab sampling. This difference was statistically significant with a p-value of less than 0.01.

Cryptosporidium oocyst log removal results were compared with Giardia cyst removal results based on continuous sampling from all six experiments. Giardia cysts were more readily removed than Cryptosporidium oocyst (p-value less than 0.01). The mean log Giardia cyst removal was 0.7 log higher that the mean log removal for Cryptosporidium oocysts. The higher level of Giardia cyst removal may be due to their larger size.