Methods for assessing long-term mean pathogen count in drinking water and risk management implications.

Area of Science:

• Environmental microbiology
• Water quality assessment
• Statistical modeling

Background:

• Pathogen counts in water often follow discrete Weibull (DW) or discrete growth distributions (DGD).
• These distributions are highly skewed, making it difficult to estimate rare high-count events and the long-term mean.
• Existing datasets may not adequately represent these critical high-count events.

Purpose of the Study:

• To evaluate the methods and data record lengths needed for assessing long-term mean microbial counts.
• To analyze microbial count data for correlations and cycles using spectral analysis.
• To provide recommendations for water quality management strategies.

Main Methods:

• Simulated representative DW and DGD waterborne pathogen count distributions.
• Spectral analysis of microbial count data to identify correlations and cycles.
• Evaluation of data record lengths required for reliable mean estimation.

Main Results:

• Longer data records are generally needed for more highly skewed distributions, typical of highly treated water.
• Approximately 500-1,000 random samples are required for reliable assessment of the population mean (±10%).
• 50-100 samples can provide an estimate within one log (45%) below the mean.
• A first-order model generated count series with a 1/f signal, observed in empirical data.

Conclusions:

• Accurate assessment of long-term microbial means necessitates substantial data, particularly for skewed distributions.
• Understanding data correlations and cycles is crucial for effective water sampling strategies.
• A tiered management approach, including rapid response to indicator deviations, is recommended for water safety.