EXTERNAL TECHNICAL REPORT Updating and revising the historical datasets in EFSA s Zoonoses databases in Finland Rintakoski S., Tuominen P., Raulo S., Mikkelä A. & Savela K. Finnish Food Safety Authority Evira Abstract EFSA has an important role to collect an annual reporting data on zoonosis, zoonotic agents, animal population, and antimicrobial resistance in the EU in accordance with the Directive 2003/99 /EC. EFSA is also harmonizing and developing the electronic data transmission on this bio monitoring field and has published the data models and standard terminologies for aggregated data collection. From 2012 the Member States and other reporting countries have been able to submit data according these standards and rules to EFSA S. EFSA also supports the Member Countries in this transition phase and has awarded projects with grants. Also the Finnish Food Safety Authority Evira has been awarded with a grant. The main objective of this project was to validate prevalence, disease status and animal population data for zoonotic agents in Finland between 2004 and 2012. All of these three tables are now validated and data has been uploaded to EFSA DCF. Project also aimed to implement new improved methods for data collection and data input according to results and knowledge gained from the data validation. We suggest that a key variable will be added to the pick list in order to easily separate duplicate reporting for events that have more than one positive zoonotic species. This project works with close collaboration with the IT development project GP/EFSA/BIOMO/2012/02- Implementation and testing on electronic submission in XLM, Excel and CVS formats of monitoring data on zoonoses and aims to improve the whole national process regarding the monitoring of zoonoses. Validation of the historical data also serves national risk assessment and epidemiological research. This document is the Final report of the project OC/EFSA/BIOMO/2013/01 and describes the validation process and findings in the project. Finnish Food Safety Authority Evira Key words: Zoonosis, data evaluation, monitoring, historical data Question number: EFSA-Q-2014-00631 OC/EFSA/BIOMO/2013/01 Correspondence: data.admin@efsa.europa.eu Disclaimer: The present document has been produced and adopted by the bodies identified above as author(s). This task has been carried out exclusively by the author(s) in the context of a contract between the European Food Safety Authority and the author(s), awarded following a tender procedure. The present document is published complying with the transparency principle to which the Authority is subject. It may not be considered as an output adopted by the Authority. The European Food Safety Authority reserves its rights, view and position as regards the issues addressed and the conclusions reached in the present document, without prejudice to the rights of the authors. EFSA Supporting publication 2015:EN-807
Summary EFSA has an important role to collect an annual reporting data on zoonosis, zoonotic agents, animal population, and antimicrobial resistance in the EU in accordance with the Directive 2003/99 /EC. EFSA is also harmonizing and developing the electronic data transmission on this bio monitoring field and has published the data models and standard terminologies for aggregated data collection. From 2012 the Member States and other reporting countries have been able to submit data according these standards and rules to EFSA S. EFSA also supports the Member Countries in this transition phase and has awarded projects with grants. Also the Finnish Food Safety Authority Evira has been awarded with a grant. The main objective of this project was to validate prevalence, disease status and animal population data for zoonotic agents in Finland between 2004 and 2012. All of these three tables are now validated and data has been uploaded to DCF data warehouse. Project also aimed to implement new improved methods for data collection and data input according to results and knowledge gained from the data validation. We suggest that a key variable will be added to the pick list in order to easily separate duplicate reporting for events that have more than one positive zoonotic species. This project works with close collaboration with the IT development project GP/EFSA/BIOMO/2012/02- Implementation and testing on electronic submission in XLM, Excel and CVS formats of monitoring data on zoonoses and aims to improve the whole national process regarding the monitoring of zoonoses. Validation of the historical data also serves national risk assessment and epidemiological research. This document is the Final report of the project OC/EFSA/BIOMO/2013/01 and describes the validation process and findings in the project.
Table of contents Abstract... 1 Summary... 2 Table of contents... 3 1. Introduction... 4 1.1. Terms of Reference as provided by the requestor... 4 2. Data and Methodologies... 4 2.1. Data... 5 2.2. Methodologies... 5 3. Assessment... 5 4. Conclusions... 7 5. Recommendations... 8 6. Glossary and Abbreviations... 8
1. Introduction National representatives and international organisations assist EFSA by gathering and sharing information on zoonoses in their respective countries. This data is then used to monitor and analyse the situation of zoonoses, antimicrobial resistance and food-borne outbreaks across Europe. This information has originally been collected with web-based forms but is has in 2013 switched to electronic data collection with pre-defined data input fields to gain better use of the data. In Finland, several different units inside Evira have been responsible for reporting data to EFSA in previous years. Lack of centrally organised data input the process has led to respective amount of inconsistent reporting and transfer error. This project aimed specifically to validate the historical data and identify and correct the erroneous information found in the data. Although food-borne outbreaks and antimicrobial resistance are part of EFSA s data monitoring aims there was currently no need for updating these in Finland. Finland is part of The Task Force on Zoonoses Data Collection and has been reporting data since 2004. Although data on zoonoses has been collected, Finland does not have national program to analyse and report yearly statistics of the zoonoses. Finland has therefore been dependent of the EFSAs yearly reports. 1.1. Terms of Reference as provided by the requestor This contract/grant was awarded by EFSA to: Kirsti Savela Finnish Food Safety Authority Risk Assessment Unit Mustialankatu 3 00790 Helsinki FINLAND Contract title: Updating and revising the historical datasets in EFSA s Zoonoses databases Contract number: OC/EFSA/BIOMO/2013/01 LOT2 2. Data and Methodologies In Finland, the data collection had not been thoroughly organized on previous years and was therefore in need of validation. The evaluation and correction of the historical zoonoses data started in January 2014 by requesting all Finnish zoonotic data from EFSA. The data collection and stage one of the data analyses started in April 2014. Finnish data was evaluated and errors in the data were corrected between April 2014 and March 2015. In addition the definitions of zoonotic agents were harmonized according to EFSA s pick list for international use.
2.1. Data Correction of Finnish historical zoonosis data (2004-2012) 2.2. Methodologies The validation focused on zoonotic prevalence tables and specifically on mandatory variables that are required by EFSA for the yearly reporting. First stage was to look the data for possible duplicates and remove them. Second stage was to quantify how much of the mandatory variables had missing information. Thirdly validate the existing observations that all the observations in the final data were correctly representing the Finnish prevalence information in each of the mandatory variables. Project also validated the disease status tables between years 2005 2012 and population tables between years 2004-2012. Similar process was used for disease status and animal population tables in the validation process. Harmonized coding for zoonotic agents (EFSA): To provide comparable information between the countries the data must validated but also have uniform coding for all observations. For uniform coding EFSA provides data model and pick list tables. All of the observations in animal population, disease status and prevalence tables were modified according to EFSA pick list information. After the validation and re-coding of the variables to meet the EFSA standards the data tables were coded to SSD codes and were transmitted to EFSA s data warehouse using the Data Collection Framework. 3. Assessment oporiginal raw data in the prevalence tables between 2004 and 2012 had 7827 observations. After the data validation the number of observations in the prevalence data was 1607 observations. For both disease status data and the animal population data the validation was significantly easier. For Disease status data (2005-2012) there was only one major change in the matrix coding were farmed deer had wrong coding. On animal population data (2004-2012) there were 666 observations of which 98 were removed due to duplicate information or missing population information leaving 574 observations in total. Duplicate information Most of the observations in the prevalence tables were duplicates. After removing duplicates the prevalence table had 1686 observation. The reason for large amount of duplicate information is not clear but the most likely explanation is that the prevalence database had large amount of ready filled fields for the matrix variable that were updated each year. In case some of the matrix fields did not have any test that current year observation was still reported to EFSA as zero tested and zero positive. These zero test results constituted the vast majority of the duplicates found in the prevalence data. In addition to these duplicates also 77 duplicates were identified and removed. These cases were also zero observation but with distinction that the same test matrix had also positive findings for other zoonosis. Example illustrated in table 1.
Table 1. Three Campylobacter observations in the prevalence data where the last of the observations (C.lari) is non-informative observation and removed from the prevalence table. The only identified Campylobacter species were C. coli and C. jejuni. no C. lari species were identified and therefor were removed. totunits totunits units matrix zoonosis Tested Positive Positive Gallus gallus (fowl) - broilers Campylobacter - C. coli 1315 82 8 Gallus gallus (fowl) - broilers Campylobacter - C. jejuni 1315 82 74 Gallus gallus (fowl) - broilers Campylobacter - C. lari 1315 82 0 Validation of data After the removal of the duplicate information the concentration was on validation of the existing data. Mainly mandatory variables were evaluated due to time constrains. Table 2 shows all of the variables and how many observation of each of the variables were corrected. Over 90% of all corrected observations were missing observations. Due to the nature of the prevalence tables all of the observations in the validations process needed to be manually checked and corrected. Most of the over 8000 misclassified observations were individually checked and studied in order to ensure correct value or classification.
Table 2. Variables in the prevalence table. Mandatory of the variable and amount of observations that needed correction on each variable Variable Variable name Mandatory corrected obs (tot. n=1607) PRV.01 Reporting year repyear Yes 0 PRV.02 Reporting country repcountry Yes 0 PRV.03 Language lang Yes 0 PRV.04 Zoonotic agent zoonosis Yes 79 PRV.05 Matrix matrix Yes 58 PRV.06 Sampling stage sampstage Yes 558 PRV.07 Sample origin samporig No 1016 PRV.08 Sample type samptype No 1140 PRV.09 Sampling context sampcontext Yes 912 PRV.10 Sampler sampler Yes 1015 PRV.11 Sampling strategy progsampstrategy Yes 1093 PRV.12 Sampling details sampdetails No - PRV.13 Area of sampling samparea No 1593 PRV.14 Sampling unit sampunit Yes 83 PRV.15 Sample weight sampweight No Only Listeria59 PRV.16 Sample weight unit sampweightunit No Only Listeria 59 PRV.17 Source of information sourceinfo No - PRV.18 Target verification target No - PRV.19 Number of flocks under control programme contrflocks No - Only coxiella PRV.20 Number of clinically affected herds affectherds No 3 PRV.29 Vaccination status vaccination No - Only Listeria 55 PRV.21 Total units tested totunitstested Yes PRV.22 Total units positive totunitspositive Yes 0 PRV.23 Analytical method anmethcode No PRV.24 Quantity quantity No Only Listeria 68 Only Listeria 83 PRV.25 Number of units tested unitstested No PRV.26 Number of units positive unitspositive No 258 PRV.27 Comment rescomm No 4. Conclusions
This project provided accurate and validated historical zoonoses data for both national and EU level. The results also suggest that some changes are essential in Finnish data collection methods to ensure that data collection in the future meets the ESFA s standards and that all mandatory information is reported. These changes are already being implemented in Finland during 2014 reporting in Finland. Project also improved the national monitoring by providing accurate long time data for zoonotic agents for better prevalence estimation in Finland. Results also provided data that is now comparable among other member states that have validated their historical data. Historical data can now also be used in research purposes. To calculate prevalence from the current data format one has to take into account the data structure which allows multiple observation for the same test types. This occurs on events such as Salmonella where more than one positive serotype is found from the test. In order to easily separate these duplicates for prevalence calculations or data management we suggest that new key variables will be added to the EFSA pick list. This variable would be unique for every new test types but duplicate for those observation that have found multiple positive zoonoses (such as different Salmonella serotypes). This would make unique test information easier to manage and also could be used as help variable in error reporting. 5. Recommendations Key variables will be added to the pick list in order to easily separate duplicate reporting for events that have more than one positive zoonotic species. 6. Glossary and Abbreviations CSV EFSA EU Evira XML Evira comma separated values European Food Safety Authority European Union Finnish Food Safety Authority Extensible Markup Language Finnish Food Safety Authority Evira