On 8 May 2013, Germany reported seven cases of hepatitis A virus (HAV) genotype IA infection in persons with a travel history to ski resorts in northern Italy. Subsequently, Italy reported an increase in the number of HAV cases at national level and declared an outbreak. At European Union (EU) level, confirmed and probable epidemic case definitions were adopted, with reference to the outbreak strain (OS) genotyping sequence result (GenBank accession number KF182323). Since 1 January 2013, 1444 cases associated with this HAV outbreak have been reported by 12 EU/European Economic Area (EEA) countries. Of these, 331 were confirmed cases. Italy reported 90% of the cases. Dispersed or clustered cases without any travel history were also reported in Finland, France, Germany, Ireland, the Netherlands, Norway and Sweden. To date no deaths associated with this outbreak have been reported; however, surveillance systems for HAV infections are not always able to capture this information.
Since August 2013, the majority of the cases in the affected Member States (MSs) have been interviewed using questionnaires adapted from the questionnaire initially developed by the Health Protection Surveillance Centre (HPSC), Ireland. At the European level, 245 completed questionnaires from confirmed cases were analysed, and 198 (80.1 %) reported consumption of berries during the relevant exposure period. In Italy, a retrospective matched case–control study identified berries as the highest associated risk factor for developing the disease (matched odds ratio (OR) 4.99; 95 % confidence interval (CI) 1.3–18.92), followed by raw seafood (matched OR 4.46; 95 % CI 1.10–18.04). A matched case–control study conducted in Ireland found that 91 % of cases had consumed at least one of four products containing frozen berries, compared with 39 % of controls (matched OR 12, 95 % CI 1.5–94). Two case–control studies were conducted in Norway. In the first study, none of the exposures was significantly associated with the disease; however, in the second study, evaluating exposure to a specific suspect cake containing berries, the matched OR was 13 (95 % CI 1.7–110). A descriptive epidemiological study in the Netherlands identified consumption of fresh soft fruit as a risk factor. Food histories from two Swedish domestic cases reported consumption of berry smoothies at a resort. It is important to note that the exposure to berries may have been underestimated, owing to recall bias caused by the long interval between exposure and onset of symptoms/diagnosis and the fact that berries may be a minor ingredient or used as a decoration in food.
In addition to the epidemiological evidence, laboratory testing identified 16 HAV-contaminated lots of frozen berries or berry products. Two lots were confirmed by sequence analyses to be contaminated with HAV OS. The intensive sampling regime for frozen berries in Italy identified 15 of the confirmed batches, i.e. batches with a positive analytical result for the presence of HAV RNA, and 45 suspect lots, i.e. lots consumed by at least one confirmed case in 2013, manufactured by 11 frozen berry packing companies. A preliminary result for one Italian lot of mixed frozen berries showed contamination with a HAV OS. In France, in February 2014, a cluster of three confirmed cases of HAV infection with an epidemiological link to a catering service was detected. Official sampling of leftovers of the suspect lot of frozen mixed berries used to produce the pastries and of the suspect lot of pastries tested positive for the presence of HAV RNA. Samples of a suspect berry mix cake in Norway were examined and found to be positive for HAV OS. Interpretation of the microbiological evidence should take account of the fact that HAV contamination is not necessarily homogeneously spread throughout a food item. This means that detection of the virus can be difficult and multiple samples need to be taken—a negative result is not sufficient to exclude the possibility of HAV contamination. In addition, genotyping of food items is more difficult than genotyping of human samples as the level of viral contamination is often too low to allow sequencing. For these reasons, HAV OS status could not be confirmed for many of the HAV-contaminated lots associated with this HAV outbreak.
The starting points for tracing were classified based on the strength of evidence (microbiological and epidemiological) of the association between food vehicle (berries) and the HAV outbreak. Therefore, evidence was rated as being of very high strength when food was confirmed to be contaminated with HAV OS or HAV-contaminated food items were known to have been consumed by confirmed cases. The strength of evidence was rated high when a suspect lot was consumed by at least one confirmed case and the exact lot and brand could be identified from the food history or when a lot had a positive analytical result without further genotyping. Possible lots, lot or lots that confirmed cases may have consumed during their exposure period (i.e. 15 to 50 days) before onset of illness, were rated as evidence of medium strength. Starting points with low strength of evidence were not traced. The tracing activities in winter 2013 started with 38 lots/cases in Italy, Ireland and the Netherlands. In spring 2014, an additional five lots/cases were added from France, Norway and Sweden. Primarily, frozen blackberries, raspberries, blueberries/bilberries and redcurrants were traced. In total, 8 starting points were classified as having very high-strength evidence, 14 as having high-strength evidence and a further 21 starting points were classified as having medium-strength evidence. The tracing data compiled for this report were collected, using a standard data format and template, by the national and regional authorities via the European Rapid Alert System for Food and Feed (RASFF). The tracing dataset comprises 6 227 transactions from 1 974 food operators (i.e. freezing processors, primary producers).
Descriptive analysis of the berry ingredients of the traced lots showed that all lots classified with very high or high strength of evidence contained blackberries, and the majority contained redcurrants. The most frequently identified origin of berries for all evidence classes was Poland or Bulgaria. Only two lots with the evidence class very high- or high- did not contain Polish redcurrants. Bulgaria was the most frequently identified origin country for blackberries, but Bulgarian blackberries were not found in all lots for any of the evidence classes. When evaluating the descriptive analysis it is important to remember that these two countries are major producers and exporters of frozen berries, while Poland is known to be the largest producer of redcurrants in Europe. In addition, there was a high level of completion of tracing data for products of Polish origin (either to the primary producer or to the freezing processor), but for other countries final end points were often missing and therefore the origin country may not be correctly identified.
The Federal Institute for Risk Assessment (BfR) in Germany has developed an open-source software tool called “FoodChain-Lab”, under a General Public License (GPL), which supports tracing back and forward analysis of suspicious food items along food supply chains. This software was used to analyse and visualise the large and complex dataset resulting from the tracing activities. Starting at the locations identified as the most probable sites of exposure, the tool allows tracing along the supply chain to identify connections between different disease clusters. This is accomplished by constructing and visualising interactive network graphs. A network graph consists of nodes and connections (edges) between the nodes. A node or station is defined as a food operator which produces, trades, stores or handles a suspect food item. The network analysis was used to identify possible “hotspots”. “Hotspots” are stations in the tracing net that are connected to a large number of starting points and therefore able to explain a large number of contaminated lots or HAV cases. A contamination event could occur at a “hotspot” or at a point prior to the “hotspot” in the supply chain.
No single point source of contamination (“hotspot”) linking all cases and contaminated lots (starting points) identified during the multinational outbreak could be determined. For redcurrants, one or more of three “hotspots” (PL#273, PL#98 and PL#115) are linked to Italian and French lots plus the Irish cases. Four common primary producers of redcurrants were identified linking these hotspots. A further two Polish freezing processors are linked to Italian lots (PL#61, PL#1237). The Norwegian lots are linked to Polish freezing processor PL#1810 and the Swedish cases to Polish freezing processor PL#260. Possible cross-contamination at stations further down the food chain (e.g. at IT#25) could link additional cases and lots to these freezing processors. Seven Polish freezing processors are linked to a total of 31 cases and lots. For blackberries, “hotspot” BG#8 plus two additional Bulgarian frozen suppliers (BG#1222 and BG#1260) are linked to Italian lots. The French lot and Irish cases are linked to the Bulgarian frozen supplier BG#1807 and BG#1884 and the Norwegian lot is linked to Bulgarian company BG#1808. Five Bulgarian suppliers of blackberries are linked to a total of 23 cases and lots. In most of the cases the tracing could not be completed to the level of primary producer or fresh berry supplier in Bulgaria. Some lots of Bulgarian blackberries comprised fruit that had been collected from public areas and forests by local pickers.
Reviewing the strongest evidence, the two lots contaminated with HAV OS contained both redcurrants and blackberries. The redcurrants were produced in Poland by two different production sites (PL#115 and PL#1810) over two harvest years (2011, 2012). The Norwegian lot contains blackberries delivered from Bulgaria (BG#1808), while the blackberries in the Italian lot were supplied by Serbia (RS#186). In addition, there is strong evidence of HAV contamination of berries at Polish freezing processor PL#273. Berry mixes from this freezing processor were used in the contaminated pastry from France and supplied to the berry product manufacturer in Ireland whose products were consumed by nine confirmed Irish cases. Suppliers of Polish redcurrants and Bulgarian blackberries were common to both the Irish and French berry mixes. Four primary producers that supplied redcurrants to PL#273 (harvest 2012) also supplied redcurrants to “hotspot” PL#98 (harvest 2011).
Overall, there are two possible conclusions. The first is that there was a single point source for the outbreak; however, information that would confirm this, relating to the sharing of resources among the berry freezing processors, e.g. exchange of temporary workers, shared use of equipment during harvest, storage or processing, a common contaminated reservoir of water for irrigation, fungicide/pesticide application or cleaning, or flooding of a wider production area, is lacking. Alternatively, a high-risk practice in the production or freezing of berries or endemic occurrence of HAV OS in the region of production or the subpopulation of seasonal workers resulted in HAV contamination in multiple locations and at multiple time points. However, several factors prevent a final conclusion being reached: uncertainties related to different investigation and sampling strategies in the affected countries; a lack of background information on berry production methods, berry products on the market and prevalence of HAV strains; the fact that only information relating to confirmed/suspect lots was analysed; recall bias on the part of cases; and the possibility of errors in the collation of the tracing data. Further local investigations are necessary to identify whether a single point source or a continuous source of contamination exists.
Based on the findings of the investigation that HAV contamination could be occurring at the freezing processor or in primary production of berries, compliance with Good Hygiene Practice (GHP), Good Manufacturing Practice (GMP) and Good Agricultural Practice (GAP) is recommended for countries producing berries for freezing. It is possible that contaminated product related to this outbreak could still be circulating in the food chain; hence, for the public health domain, enhanced surveillance, risk communication, vaccination and further research are recommended. Finally, to support future investigations and tracing activities, use of multidisciplinary teams, data exchange protocols and access to improved data management and analytical tools is recommended.