Papini Paolo, Faustini Annunziata, Manganello Rosa, Borzacchi Giancarlo, Spera Domenico, Perucci Carlo A
Agenzia di sanità pubblica della Regione Lazio, Roma.
Epidemiol Prev. 2005 Sep-Dec;29(5-6):259-63.
To determine the frequency of sampling in small water distribution systems (<5,000 inhabitants) and compare the results according to different hypotheses in bacteria distribution.
We carried out two sampling programs to monitor the water distribution system in a town in Central Italy between July and September 1992; the Poisson distribution assumption implied 4 water samples, the assumption of negative binomial distribution implied 21 samples. Coliform organisms were used as indicators of water safety. The network consisted of two pipe rings and two wells fed by the same water source. The number of summer customers varied considerably from 3,000 to 20,000.
The mean density was 2.33 coliforms/100 ml (sd= 5.29) for 21 samples and 3 coliforms/100 ml (sd= 6) for four samples. However the hypothesis of homogeneity was rejected (p-value <0.001) and the probability of II type error with the assumption of heterogeneity was higher with 4 samples (beta= 0.24) than with 21 (beta= 0.05).
For this small network, determining the samples' size according to heterogeneity hypothesis strengthens the statement that water is drinkable compared with homogeneity assumption.
确定小型供水系统（居民人数<5000）的采样频率，并根据细菌分布的不同假设比较结果。
1992年7月至9月期间，我们开展了两项采样计划，以监测意大利中部一个城镇的供水系统；泊松分布假设意味着采集4份水样，负二项分布假设意味着采集21份水样。大肠菌群被用作水安全性的指标。该供水网络由两个管环和两口由同一水源供水的水井组成。夏季用户数量在3000至20000之间变化很大。
21份水样的平均密度为2.33个大肠菌群/100毫升（标准差=5.29），4份水样的平均密度为3个大肠菌群/100毫升（标准差=6）。然而，均一性假设被拒绝（p值<0.001），在异质性假设下，4份水样的II型错误概率（β=0.24）高于21份水样（β=0.05）。
对于这个小型供水网络，根据异质性假设确定样本量，与均一性假设相比，更能有力地证明水是可饮用的。
