Reporting of foodborne and waterborne outbreaks in Finland was reorganized and intensified in 1997; new regulations emphasized that all suspected cases should be immediately reported to the National Public Health Institute (KTL). Recommendations were given for properly collecting both patient and environmental samples. The functions of local outbreak investigation teams were clarified and included training in conducting epidemiologic surveys. Laboratory performance was improved by including options for viral and protozoan diagnostics from both patient and environmental samples. All cases in which water was suspected as the source of the outbreak were reported to KTL. Sampling recommendations included 3–10 representative patient stool samples. Water samples, raw water, and when appropriate, tap water from different parts of the distribution network were collected immediately. Despite recommendations, not all outbreaks were investigated for viruses. The criteria for establishing an outbreak as waterborne were according to the English classification (grades A–D) (11).

In total, 271 patient samples from 25 outbreaks were analyzed for viruses. The range of fecal samples obtained from each outbreak was 2–69 (mean 11). A 10% fecal suspension in 0.05 mol/L Tris-HCl, 0.1 mol/L NaCl, 1 mmol/L CaCl₂, pH 7.4, was used for RNA extraction.

A total of 73 water samples from 27 outbreaks were analyzed; 1- to 2-L water samples, collected in clean glass or plastic bottles, were concentrated as described by Gilgen et al. (12). The 1-L samples were run through a positively charged disk membrane filter (diameter 47 mm, pore size 45 μm; AMF-Cuno, Zetapor, Meriden, CO, USA) with or without a fiberglass prefilter. After the elution step in 50 mmol/L glycine buffer, pH 9.5, containing 1% beef extract, the eluate was rapidly neutralized with HCl. The volume was further reduced to ≈100 μL with a microconcentrator (Centricon-100, Amicon, Beverly, MA, USA). This sample was used for RNA extraction and PCR as described (10).

RNA extraction and RT-PCR for the norovirus polymerase region were performed as described (13). Briefly, RNA was extracted by using phenol- and guanidine thiocyanate–containing Tripure reagent (Roche, Indianapolis, IN, USA) and precipitated with ethanol. Viral RNA was transcribed to cDNA, and DNA amplification was performed in separate tubes for norovirus genogroups I and II (GI and GII) by manual PCR with primers Nvp110 (14) and N69 (15), and Nvp110 and NI (16), respectively. From 2002 on, the forward primers for the genogroups were modified as KA1 (5´-GANGGCCTSCCMTCWGGNTT-3´) and KA2 (5´-TGGAATTCNATHGCCCAYTGG-3´). The amplicons were visualized by electrophoresis in an agarose gel, hybridized by a probe panel, and used for nucleotide sequencing.

Sequencing was performed manually (Sequenase, version 2.0 DNA sequencing kit, USB, Cleveland, OH, USA) as described (13). Sequence analysis was performed by programs SeqApp and ClustalW. Our sequences were aligned with the following EMBL/GenBank noroviruses: Southampton/91/UK (L07418), Norwalk/68/US (M87661), Malta (AJ277616), Melksham/94/UK (X81879), Hawaii/76/US (U07611), Lordsdale/93/UK (X86557), GIIb (AY7732101), GIId (AF312728), and murine norovirus (AY228235). For nucleotide sequences for Hillingdon/94/UK and Grimsby/95/UK, see Vinje et al (17); sequence of Lord Harris comes from the sequence database of the European network (9,18). GenBank accession numbers for nucleotide sequences of this study are AY958213–9 for GI and AY958204–12 for GII noroviruses.

As part of the improved and intensified outbreak surveillance system in Finland, we have identified waterborne viral outbreaks since 1998. In a relatively brief period, during which norovirus diagnostics have been available for patient as well as environmental samples, a considerable number of waterborne norovirus outbreaks have been detected.

That Finland has >1,300 water treatment plants may in part explain the numerous outbreaks. Many of these plants still use surface water (lakes or rivers) as raw water. Inadequate disinfection is then the most common reason for waterborne epidemics, as was the case in outbreak E1 (10). At risk also are water plants that use groundwater and no disinfection. In Finland, snow melts in spring while the ground is still frozen, which leads to surface runoffs and flooding. Breaks in sewer lines in the vicinity of a well caused several large waterborne outbreaks. Poor sewage disposal also caused many small waterborne outbreaks in private homes or rental cottages.

The large number of genetically distinct norovirus genotypes has been advantageous in investigating waterborne epidemics. Although the short amplicon sequence does not definitively show that 2 viruses are identical, for the purpose of source tracing it seems adequate. In this study, a unique viral sequence appeared in most norovirus outbreaks, and viruses from patients and water in a particular outbreak showed identical sequences. The success in most outbreaks in identifying a norovirus with the same sequence from patients and water may be due to the fact that the outbreaks have taken place in small communities. In large waterborne outbreaks, usually >1 norovirus strain and often other viruses and microbes are causative agents.

Both norovirus genogroups occurred in waterborne epidemics. In a 5-year study (1998–2002) in Finland, GII outbreaks clearly outnumbered those caused by GI noroviruses (86.9% vs. 13.1%) (21). In waterborne outbreaks, however, nearly half were caused by GI viruses. Some differences may occur in stability as well as ability to spread from person to person among viruses representing different genotypes. Type GI.3, the most common GI genotype in water samples, was also the most frequent GI type in community outbreaks (21). Viruses of this genotype have caused waterborne outbreaks in the United States in 2001 (22) and in the Netherlands (23).

As might be expected, keeping in mind its ubiquity (24,25), the GII.4 genotype was present in several waterborne outbreaks, and in Finland it has been the most frequent genotype in all outbreaks. The GII.4 new variant emerged in Finland in June 2002, and in the following year 2 waterborne outbreaks were caused by this new variant (20). Another emerging genotype, GIIb, found in Finland in 2001, a year later than in southern parts of Europe, was a causative agent in a waterborne outbreak in 2002. A waterborne outbreak in Sweden caused by this genotype has recently been reported (26).

Environmental virology of human pathogen detection has a rather limited history. A classic case is the monitoring of polioviruses in sewage (27). This method, based on a cell culture technique, is sensitive in detecting circulating wild poliovirus. Further efforts in environmental virology were lacking for many years, mainly because suitable methods were absent. Only after gene amplification techniques were introduced could a tool be developed to successfully detect norovirus in environmental samples (8,10). In recent years, an increasing number of reports have described waterborne norovirus outbreaks through contaminated drinking or recreational water (22,23,28,29).

National recommendations for volumes of water to be tested vary between tens and hundreds of liters. Such volumes pose a serious practical problem for the testing laboratory. For viral detection by RT-PCR, a smaller volume (1 L) is preferred, as suggested by Gilgen et al (12). Independent of the concentration method, the increase in RT-PCR inhibitors usually sets limits on the water concentration. Sensitive methods are needed to detect viruses in environmental samples. Recent reports on the applicability and sensitivity of real-time RT-PCR (30–32) for noroviruses also offer new possibilities to enhance its sensitivity. Another factor is that the test then becomes more rapid, which is essential in monitoring water quality, particularly in epidemic situations. The third advantage is that a quantitative estimate of the contamination level is obtained.

Microbial risks from water are recognized, with much emphasis on risk assessment (33). Assessment of water, however, depends on indicator organisms, such as coliforms or enterococci, whose survival in water is shorter than that of enteric viruses, especially norovirus and hepatitis A virus. Therefore, viruses can easily be harbored in "microbiologically immaculate" water (34,35). In situations in which a well is contaminated by sewage, coliforms are nearly always found. When sewage is released into lake water that serves as raw water downstream, indicator organisms may no longer be detectable, but noroviruses can still be present and cause illness. This sequence of events probably led to the first outbreak we examined (E1) (10).

When water plants use surface water, the contamination may be short-lived and may have vanished by the time the outbreak is detected. A "rolling sample" system might be used in which samples are collected in water plants at risk for contamination at regular intervals (e.g., daily, weekly) and stored at 4°C. Unless signs of an outbreak appear, the samples can be discarded at the same pace that new ones are collected. In case of contamination, water samples would be available for analyses.

The evidence presented here together with several recent reports mentioned above show the role of viruses as contaminants of drinking water. In Finland, the finding that noroviruses frequently cause waterborne outbreaks has led to authorities' increased awareness of viral risks. As a consequence, laboratory techniques have been improved, and the capacity for analyzing environmental samples, especially water, has increased. Legislative measures for viral monitoring as part of the microbial risk assessment in drinking water production should be seriously considered.