Should I use one sample or two samples per month for Cryptosporidium sampling?

 

Why LT2ESWTR has separate requirements for one sample per month versus two samples per month?

One of the new concepts incorporated into the LT2ESWTR is to link treatment requirements in the rule to results from Cryptosporidium source water monitoring. In developing the LT2ESWTR source water Cryptosporidium monitoring requirements, USEPA and the FACA stakeholders had to weigh two competing factors. On the one hand, collecting samples for a long time period spanning several months or years is desired so that events representing fluctuations in water quality conditions have a greater chance of being captured during the monitoring period. On the other hand, if the source water is characterized by a high Cryptosporidium risk (i.e., Bin 2, 3, or 4), this needs to be established as soon as possible so that corrective action can be initiated. Therefore, the compromise solution was to require collection of at least two years of data in order to limit the length of time prior to initiation of corrective action, while collecting sufficient data to capture events during multiple seasons (eight seasons in a two year study).

Evaluations during preparation of the LT2ESWTR indicated that the overall mean for 48 or more samples had an error rate of <2% for both false positives (facility assigned to bin level that is too high) and false negatives (facility assigned to bin level that is too low). The overall mean of 24 samples had a false negative rate that was too high (6%), but using the LT2ESWTR approved approach (highest average of any 12 consecutive months in the 24 month period) produced a more acceptable false negative error rate (<2%), though false positive rate is higher with this approach (5.3 versus 2.8 %). Therefore, the LT2ESWTR requires overall mean to be used for 48 or more samples, but requires maximum average for any 12 month period when 24-47 samples are used.

How is bin assignment determined for systems with fewer than 48 samples? How is it different for systems with 48 or more samples?

When 24-47 samples are collected, the average is determined for each month, then these monthly average values are averaged for each 12 month period. The largest of 12 month averages is used to establish bin assignment when the number of samples is less than 48. When 48 or more samples are collected, the overall average of all of these samples is used to establish bin assignment.

How would a utility benefit from collection of two samples instead of one? Doesn’t it cost more to analyze more samples?

LT2ESWTR compliance monitoring for 48 samples is expected to cost about $25,000 (including filters), versus $13,000 for 24 samples. Therefore, analyzing two samples a month for 24 months will cost about $12,000 more than analyzing only one sample per month. However, most utilities would probably consider this cost difference a worthwhile investment given the consequences of being improperly assigned to a bin level that is too high (false positive). Using LT2ESWTR information on false positive and negative error rates, for the $12,000 investment of increasing from 24 to 48 total samples, a utility would marginally improve the false negative error rate (1.7 to 1.4%), but would markedly improve the false positive error rate (5.3 to 1.7%). Consequently, a utility could potentially avoid unnecessary expenditures by collecting enough samples to make sure they are assigned to the appropriate bin.