Cryptosporidium and Giardia Occurrence in Icr Drinking Water Sources—statistical Analyses of Icr Data

Because of low occurrence concentrations, small sample volumes, and low and variable recoveries associated with analytical methods, most measurements of Cryptosporidium and Giardia in water samples collected under the Information Collection Rule (ICR) were negative; many locations produced nothing but zero results over the duration of ICR monitoring. The greater median volumes analyzed and the higher recoveries achieved with the method used in the Supplemental Surveys (Method 1622) yielded fewer zeros, but some locations still showed no positive counts. Bayesian statistical techniques were used to estimate national Cryptosporidium and Giardia occurrence distributions that were consistent with measured results. The statistical model accounted for the volumes analyzed and the low but variable recoveries. Median estimates were produced for the national distribution of mean concentrations of Cryptosporidium and Giardia in water treatment plant influent. Uncertainties in these estimates were represented by 90% credible bands. These same techniques were used with data sets from the Supplemental Surveys to estimate occurrence in large and medium-sized systems. Distributions of Cryptosporidium occurrence derived from the two Supplemental Surveys and the ICR had medians near 0.02 oocysts/L, but the ICR survey yielded a greater location-to-location variance and predicted the greatest number of water sources with very high (more than 1 oocyst/L) and very low (less than 0.001 oocyst/L) occurrence.