Realizing Official Statistics on Accidents at Work in Norway

Transcription

1 Journal of Statistical Science and Application, February 2016, Vol. 4, No , doi: / X/ D DAV I D PUBLISHING Jensen, Arne Cand.polit Political Science Statistics Norway, Division for Health Statistics Akersveien 26 NO-0033 Oslo, Norway arne.jensen@ssb.no Bråthen, Remy Cand.polit Political Science Statistics Norway, Division for Health Statistics Akersveien 26 NO-0033 Oslo, Norway remy.brathen@ssb.no Statistics Norway has been engaged in the development of official statistics on accidents at work for the last ten years and represents Norway in international bodies like Eurostat working groups. Some of the work was documented and presented back in 2011 at the ISI Dublin convention and a review of further developments the last four years could shed even more light over the efforts made. There has been implemented a new data collection system at the national level that involves data and files based on forms for reporting accidents at work being sent from the Norwegian Labour and Welfare Administration (NLW) to Statistics Norway. Nevertheless there are still some challenges to be met. These include the use of different versions of the NLW forms, the scanning and extraction of data in the NLW, the implementation of a secure electronic solution for transmitting data between the NLW and Statistics Norway, the reading and interpretation of tiff-files and the lessons to be learned from other countries. The ambition is that Statistics Norway produces methodological sound official statistics on accidents at work within the first half of 2015 and transmits data and files to Eurostat that are necessary and sufficient to fulfil EU regulations within the first half of Keywords: international reports, EU regulations, data collection system, statistical challenges. Introduction At the 58 th World Statistics Congress ISI 2011 in Dublin Statistics Norway presented the paper Developing Official Statistics on Accidents at Work in Norway. The paper described the situation in Norway regarding the data collection, different sources of basic data, and the systemic and statistical challenges Norway faced when developing official statistics on accidents at work. Furthermore, the paper introduced the different Corresponding author: Arne Jensen, Cand.polit Political Science, Statistics Norway, Division for Health Statistics Akersveien 26, NO-0033 Oslo, Norway. arne.jensen@ssb.no.

2 49 actions that were to be carried out at the national level to fulfil international obligations as well as national needs for official statistics. The situation regarding statistics on accidents at work in Norway has developed since There has been carried out a number of actions to achieve progress and subsequently produce official statistics on this statistical area. The most significant achievement has been the establishing of a new data collection system for non-fatal accidents at work between the Norwegian Labour and Welfare Administration (NLW), who is the agency handling data from the National Insurance System (NIS), and Statistics Norway, who will produce official statistics and international reports in the years to come. With regard to fatal accidents at work Statistics Norway has established data transmission from the relevant inspection authorities Despite the fact that the data collection system at this writing has not produced official statistics nor international reports, the time frame shows that official statistics is scheduled for spring 2015 while international reports according to EU regulations are set for spring/summer Furthermore, Statistics Norway has looked abroad to gather experience and search for best practice at the same time as actions has been carried out at the national level. It is a fact that differences in national social security systems and national schemes for occupational injury insurance represent challenges when it comes to comparisons across borders. Nevertheless, dialogue with countries that have related systems and schemes and that produce methodological sound statistics that complies with international regulations, are by all means useful. Therefore Statistics Norway has over the last 3-4 years been in contact with Statistics Finland regarding their official statistics and international reports on accidents at work. The paper describes both the major actions that are carried out over the last years and the most important challenges Norway has faced when it comes to realizing official statistics on accidents at work and basic data on non-fatal accidents. The actions could be summed up as implementing a new data collection system, employing Eyes and Hands software, extracting data from administrative systems, developing statistical weights and comparing with other countries How did the Implementation of a New Data Collection System Work out? At the national level there is an understanding that the best case scenario for collecting methodological sound basic data regarding accidents at work (as well as occupational diseases) is an electronic solution for reporting accidents and diseases. In practice, this means that authorized representatives for the employer, the company health service or the employee herself would be able to use an online internet service to report accidents at work (or occupational diseases) directly to The Norwegian Labour and Welfare Administration (NLW). However, an electronic solution calls for investments and developments in the NLW that are not feasible in the short run. Therefore, there is established a solution that, in short, includes that the NLW receive and scan the paper based forms and send image files (tiff-files) of the forms to Statistics Norway along with some other data extracted from the NLW registers (csv-files). In general, the main parts of the data collection system that are sketched out in figure 1 are realized by January However, there are still some challenges and unanswered questions regarding all the different stages in the process. (1) The NLW forms that are used to report accidents at work are not always the latest versions and therefore not always compatible with the use of software in Statistics Norway; (2) the procedures regarding scanning the forms are not that transparent and some questions could be raised when it comes to the data that are extracted from NLW registers; (3) the transmission of data and files from the NLW to Statistics

3 50 Norway is still done by CD/DVD and not by the envisioned SFTP-solution; (4) the reading and interpreting of tiff-files in Statistics Norway by Eyes and Hands software is not straight forward; (5) the administration of registers and the production of official statistics and international reports are not completely tested. 1. The employer, the employee or other authorized person sends the form to the NLW 2. The NLW receives and scans the form and extracts some data from their systems 3. The NLW sends scanned forms (tiff-files) and other data (csv-files) to Statistics Norway 4. Statistics Norway receives, interprets and registers information in designated registers 5. Statistics Norway administer the registers, produce official statistics and produce reports to Eurostat Figure 1. Dataflow and data collection system between the Norwegian Labour and Welfare Administration (NLW) and Statistics Norway How do we deal with Different Versions of the NLW forms? Prior to the implementation of a new data collection system Statistics Norway collaborated with The NLW to adjust and revise the contents and design of the forms used to report accidents at work. The purpose was to include information that is needed to transmit sufficient data and statistics on accidents at work to fulfil relevant EU regulations. This meant that some questions and information boxes were added to the form from 1 January 2014, e.g. the number of days the employee supposedly would be absent from work and more details on the causes and circumstances of the accident. Additionally, the purpose was to discuss how the forms had to be designed to make them readable by software used by Statistics Norway (Eyes and Hands). In general, the forms that were implemented from 1 January 2014 are read and interpreted as they should. The problems regarding the new forms are for most parts not caused by the forms themselves but the fact that employers and others reporting accidents at work use technical devices (e.g. printers) with settings and formats that cause the Eyes and Hands software to fail reading the forms. Furthermore, some reporters of accidents fill out the forms by pen which causes some worries regarding the interpretation of the hand writing. Still, this is not by far as big a challenge as the use of forms that were in use prior to 1 January The software is not set up for the use of old forms which in practice means that approximately two thirds of the forms have to be dealt with otherwise than by automatically reading and interpreting. During the first half of 2015 these forms will among other things be analyzed by means of methodological weighting and the use of register data. How do we get a Better Overview of the Scanning and Extraction of Data in the NLW? The NLW plays a major part when it comes to facilitating the data flow and produce methodological sound basic data concerning accidents at work for Statistics Norway to base the production of official statistics and international reports on. The NLW has outsourced all its scanning activities and the company which carry out the scanning of the forms producing data on accidents at work receive about forms regarding

4 51 different pensions and benefits every day. In addition to this source of information the NLW provides data for Statistics Norway that is extracted from their own administrative systems. These data contains a few variables like a personal identity number and the VAT number of the employer. The dialogue between the NLW and Statistics Norway is important in respect of extracting the correct data about the correct cases of accidents at work. The scanning of the forms is not regarded as a potential source of error as much as the filters used by the NLW when they do the actual extraction of cases that are transmitted monthly to Statistics Norway. The NLW filters which cases that are chosen based on criteria that have to be decided on in collaboration with Statistics Norway. By the time of the dissemination of official statistics in June or July 2015 the underlying criteria for filtering cases should be defined and discussed. How do we Establish Automatic data Transmission by a SFTP-solution? The NLW started the transmission of data on accidents at work to Statistics Norway back in May Except for the first delivery which contained data from January to March all but one of the transmissions has contained data for one month. More specifically the monthly deliveries include two kinds of files, tiff-files and csv-files. The first files are in fact picture files and consist of the scanned paper forms, the latter are files that include information extracted from the registers and administrative systems in the NLW. The files are delivered by DVD/CD from the NLW. Given the priorities and the uncertain time frame of the NLW when it comes to implementing electronic forms for reporting accidents at work in the short run the ambition with the new data collection system is to make sure that most of the dataflow between the NLW and Statistics Norway goes automatically. There is an agreement to implement a SFTP-solution (Secure Shell File Transfer Protocol) which facilitates secure transmission of files in an insecure network. The necessary infrastructure is established in Statistics Norway and the NLW has planned for the solution. The dialogue between the NLW and their superiors at the ministry on the one hand and Statistics Norway on the other should get the SFTP-solution implemented within a few months. How do we Make up for the Challenges Concerning Reading and Interpreting Tiff-files? Statistics Norway has set up an IT infrastructure for receiving, reading and interpreting files, and per February 2015 efforts have been made to create a system for producing official statistics at the national level and reports at the international level. In general, every piece of the puzzle and every part of the data collection and the production system is either developed or under development. One important part of the production system that should be mentioned explicitly is the Eyes and Hands software that has been implemented and adjusted to read information from the tiff-files. Another part that will prove important in the time to come is the file structure that has been set up within Statistics Norway to organize and share the data that is interpreted. With regard to the tiff-files there are a few challenges concerning the new and old forms respectively. Despite the fact that the forms produced from 2014 and onwards are designed to ease the reading and interpretation there are still some cases that Eyes and Hands are not able to read and verify. This is mostly due to the fact that there are different printers in use, which in turn creates problems when read by Eyes and Hands. Statistics Norway is pr. February 2015 in dialogue with the producers of Eyes and Hands to explore the potential of implementing versions that could cope with the problem. When it comes to forms produced before 2014 it s a different story. These forms are not set up to be read and interpreted by Eyes and Hands and have to

5 52 be dealt with by means of other actions. Pr. February 2015 efforts were made to use data extracted from the administrative systems of the NLW (csv-files) and methodological tools like weighting to estimate some of the information regarding the accidents at work. How do we Secure Methodological Sound Statistics and International Reports? Statistics Norway has implemented a data collection system that should provide basic data that are necessary and sufficient to produce methodological sound official statistics at the national level that fulfil the Code of Practice and international reports regarding accidents at work that meet the requirements in terms of EU regulations. Additionally Statistics Norway participates in working groups lead by Eurostat, the statistical office of the EU, to secure that statistical developments are carried out within the right framework. Furthermore, Statistics Norway and Statistics Finland have developed a bilateral cooperation that has proved useful and given both practical and methodological input to how the Norwegian reports on accidents at work could be improved. There are definitely experiences to be shared internationally across borders, either through Eurostat or by means of bilateral collaboration between countries, when it comes to statistical developments on the area of accidents at work. However, there is reason to believe that some of the differences between countries when it comes to national insurance systems, e.g. occupational disability schemes, are important to bear in mind when for instance data collection systems are evaluated. Among other things the fact is that the system of reporting accidents at work in Finland gives the employer a much stronger financial incentive to register accidents than in Norway. In short the Finnish system is based on an employer s insurance while the Norwegian counterpart is based on an employee s insurance. In practice this means that there is reason to believe that the reporter of accidents at work in Finland is more prone to report detailed and sound data and consequently that the quality of basic data is better in Finland than in Norway. Conclusions Statistics Norway are set to produce official statistics on accidents at work within the first half of 2015 and transmit data on this statistical area to Eurostat according to EU regulations within the first half of However, as discussed above there are still some questions to be asked and efforts to be made to meet the challenges both when it comes to official statistics and international reports. The most important questions to answer are those regarding the basic data and missing values. The production processes in Statistics Norway have to deal with the fact that a large proportion of the basic data received from the NLW is not automatically readable and therefore that other opportunities should be explored to make use of the information that pro tempore is lost. In fact, Statistics Norway is fully aware of the fact that realizing official statistics on accidents at work in Norway to a large extent depends on the quality of basic data and the tools used to extract and produce the data.