Comparing Data from the Poisons Information Centre with Employers’ Accident Reports Reveal Under-Recognized Hazards at the Workplace

Comparing Data from the Poisons Information Centre with Employers’ Accident Reports Reveal... Abstract Records of injuries and incidents provide an important basis for injury prevention related to hazardous substances at the workplace. The present study aimed to review available data on injuries and incidents involving hazardous substances and investigate how data from the Poisons Information Centre could complement the records of the Swedish Work Environment Authority. We found two major obstacles for using injury/incident data based on employers’ mandatory reporting. First, it was not possible to quickly and reliably identify injuries caused by hazardous substances, and second, data identifying substances or products are not systematically included. For two out of five investigated injuries with lost working days likely due to chemical injuries, we could not identify substances and/or products involved. The records based on calls to the Poisons Information Centre allow better understanding of chemical hazards and products. Besides the large share of unidentified chemical hazards in the injury statistics, the most striking difference was found for cleaning agents. Cleaning agents were implicated in one-third of the occupational cases that the consulting Poisons Information Centre expert judged to pose a major risk and in need of immediate healthcare. Only one in 10 injuries with lost days reported by employers was related to this type of product. The identification of exposures and symptoms by the Poisons Information Centre allow recognition of chemicals with problematic occupational uses. Hence, these records may serve as an important complement to official injury statistics related to incidents with hazardous substances at work. chemicals management, occupational health and safety, poison control, risk assessment, work injury Introduction Hazardous substances in the workplace cause a wide variety of occupational injuries and diseases. A part of the authority risk management regime is to compile information on occupational diseases, injuries and severe incidents. Analysis of this information makes up part of the authorities’ work to identify areas of interest for inspection or information efforts (Jacinto and Aspinwall, 2004). However, we have reasons to believe that occupational injury and disease statistics are incomplete and underestimate the number of injuries. Certain groups of workers may be undercounted because they fall outside the reporting requirements, such as trainees, self-employed or from the informal sector. Studies from the United States have estimated that up to two-thirds of occupational injuries are not included in the national surveillance system (Leigh et al., 2004; Rosenman et al., 2006). Under-reporting is also a cause of incomplete injury statistics. Employers may lack awareness of the legal requirements to report injuries and incidents, reporting responsibilities may be confused due to the (increasingly common) complex employment structures in many industrial sectors (Rappin et al., 2016). It is also plausible that incentives are low to report and incriminate oneself or to make the effort in the face of limited time. Furthermore, under-reporting may be due to employees’ under-reporting to their employers, for instance due to disciplinary programs or incentive programs that penalize accidents (US GAO, 2012; Fagan and Hodgson, 2017). Moreover, any underestimate is likely skewed. For instance, the degree of under-reporting has been found to be larger for less severe injuries (Gravseth et al., 2003). In addition, job insecurity has been found to correlate with both an increased risk for injury and a decreased reporting of injury (Probst et al., 2013; Picchio and van Ours, 2017). Undercounting and under-reporting of the number of chemical incidents and injuries reduces the ability to identify and address occupational safety issues with chemicals. One implication of skewed underestimates is a skewed decision-basis for the authorities, which may further reduce efficiency of their risk management measures. Additional data sources, which cover a larger or a different subset of chemical injuries and incidents, may thus prove to be valuable complements to official injury data. Data from Poisons Information Centres (PIC) have been used to provide overviews of accidents with hazardous substances at work (Blanc and Olson, 1986; Litovitz et al., 1993; Schenk et al., 2018). PICs inform the public, workplaces and/or healthcare personnel regarding the toxicity of exposure agents and the management of exposed patients. In performing these tasks, the PICs gather data regarding exposures that may be useful for managing the incident or for surveillance and follow-up. Although not a standard tool for occupational injury surveillance, such data have been used to investigate a variety of occupational issues or groups, such as exposure to pesticides (Olson et al., 1991; Meulenbelt and de Vries, 1997; Sudakin and Power, 2007; Trueblood and Shipp, 2017), workers in small enterprises (Hinnen et al., 1994) or adolescent workers (Rubenstein and Bresnitz, 2001; Woolf et al., 2001; Ziqubu-Page and Forrester, 2016). A recent overview of Swedish PIC data on occupational cases indicated that these might cover a larger range of occupational injuries or incidents involving hazardous substances than registered by the national occupational injury statistics (Schenk et al., 2018). The objective of the present work was to compare the injuries and incidents involving hazardous substances recorded under Swedish occupational injury data and in the call records of the national PIC. Specific questions addressed were: • Do the different databases cover a largely different set of injuries and incidents? • Are there differences in the patterns of chemical hazards or product groups encountered in these databases? • What are the strengths and limitations of the respective databases for investigating accidents/incidents with hazardous substances? Through addressing these questions we aimed to evaluate if and to what extent data from the PIC provide complementary knowledge that can be used for prevention, for instance by identifying hazards or chemical products likely to be under-recognized in the occupational injury data. Methods In the following sections, we describe how data were retrieved, managed and analysed. Ethical vetting was applied for and approved by the regional ethical board in Stockholm (2015/1-31/5). Ethical vetting was not sought for matching individuals case by case between data sources. Data sources Table 1 outlines the three data sources. Two of the databases are managed by the Swedish Work Environment Authority (SWEA), which provided the data excerpts used in the present study. The national statistics are compiled in a database named ISA (acronym for information system about occupational injuries). This database is based on employers’ reporting under the Swedish Social Insurance Code (Socialförsäkringsbalk 2010:110), which requires employers to report work-related injuries. If such injuries involve one or more days of absence from work, they are registered in ISA. Parts of the ISA database are publicly available online (http://webbstat.av.se/; last accessed 27 September 2017). Reporting to ISA follows a structured notification form including variables defined by SWEA and the European Statistics on Accidents at Work (ESAW; Eurostat, 2013). ESAW was under implementation during the period of the present study (Commission Directive 349/2011), when referring to ESAW variables the period of applicability will be specified. Since December 2011, ISA reports can be submitted online (https://anmalarbetsskada.se/, last accessed 27 September 2017). Also, the occupational injury benefit scheme was changed for a large part of the Swedish workforce in April 2012, with injury compensation starting from the first day of sickness absence, compared to previously from the second day of absence. This would provide an incentive for the employees to require documentation of occupational injuries under the Social Insurance Code (i.e. ISA) and possibly increase likelihood for absence from work after an injury (SWEA, 2013). Table 1. Overview of databases. ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 aAll injuries should be reported under the social insurance legislation, only >1 lost days are registered in ISA. bSevere injury is not defined, se text for examples given by SWEA of severe injuries. cAll categories at this level were ‘electricity’, ‘fire’, ‘explosion’, ‘chemical’, ‘falling or flying objects’, ‘fall of person’, ‘animals’, ‘vehicles’, ‘physical overload or misstep’, ‘machine or transport device’, hand-held tool or object’(includes syringes), ‘threat’, ‘robbery’, ‘violence’, ‘psychological chock’ and ‘other cause’. dAll categories at this level (accident was caused by) were ‘dusts, gases, fumes or liquids’, ‘electricity, explosion or fire’, ‘material fell, plummeted, burst or exploded’, ‘collision between vehicles, hit by vehicle, injured by object, machine, tool or animal’, ‘person fell’, ‘struck against something, stepped on sharp item’, ‘misstep, lift or other overload’, ‘threats or violence, chock, fear’ and ‘other cause’. These categories are unaffected by the implementation of the ESAW methodology. View Large Table 1. Overview of databases. ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 aAll injuries should be reported under the social insurance legislation, only >1 lost days are registered in ISA. bSevere injury is not defined, se text for examples given by SWEA of severe injuries. cAll categories at this level were ‘electricity’, ‘fire’, ‘explosion’, ‘chemical’, ‘falling or flying objects’, ‘fall of person’, ‘animals’, ‘vehicles’, ‘physical overload or misstep’, ‘machine or transport device’, hand-held tool or object’(includes syringes), ‘threat’, ‘robbery’, ‘violence’, ‘psychological chock’ and ‘other cause’. dAll categories at this level (accident was caused by) were ‘dusts, gases, fumes or liquids’, ‘electricity, explosion or fire’, ‘material fell, plummeted, burst or exploded’, ‘collision between vehicles, hit by vehicle, injured by object, machine, tool or animal’, ‘person fell’, ‘struck against something, stepped on sharp item’, ‘misstep, lift or other overload’, ‘threats or violence, chock, fear’ and ‘other cause’. These categories are unaffected by the implementation of the ESAW methodology. View Large In addition to ISA reporting, severe injuries and severe incidents are also to be reported separately to the SWEA under the Swedish Work Environment Act (Arbetsmiljölag 1977:1160). There is no strict definition of severe injury; injuries to internal organs, burns larger than 5% of the body surface, and accidents where several persons were injured are given as examples. Severe incidents are defined as near misses that could have caused severe injuries. These severe accident/incident records (henceforth, SAIR) are not publicly available, but are used by the SWEA in planning inspection efforts. In 2014, 13% of first time workplace inspections were in response to SAIR reports (SWEA, 2015). The SAIR notification form is shorter than that of ISA and uses different categories for cause (Table 1); electronic reporting was launched at the same date and through the same webpage as ISA. In Sweden, one PIC unit serves the whole country (population of 10 million). The 24-h phone service is open to medical professionals as well as the general public. Telephone consultations are logged in the PIC database and multiple calls about the same case are connected to the first call allowing identification of individual cases. The Swedish PIC is responsible for receiving information relating to composition of hazardous products under the EU Regulation (EC) 1272/2008 on classification, labelling and packaging (CLP). Hence, PIC experts have access to a database on product composition, aiding in the identification of major hazards when provided with the product name. Based on reported exposures (extent and agent) and symptoms, the PIC expert assesses the risk to the exposed. Minor risk cases present with no or minor symptoms and are manageable on site. Moderate risk cases may present pronounced and/or prolonged symptoms, immediate healthcare may be recommended by the PIC expert. Major risk cases present or may present severe symptoms and immediate healthcare is recommended. If the PIC expert is unable to evaluate risk, for example due to limited information, this is logged as undetermined risk. If applicable, risk estimates are replaced by confirmed severity corresponding to poisoning severity score (PSS) 3-severe injury or 4-fatality (Persson et al., 1998). Extraction of relevant cases Information system about occupational injuries The cause-of-injury category ‘dust, gases, fumes or liquids’ was used as the most relevant category for exposures to hazardous substances, yielding 1951 cases for 2010 through 2014 (Table 1). As these cases may include non-poisoning hazards, for example burns by boiling water, the records needed additional filtering (Fig. 1). Based on injury data, 594 cases were included, that is ‘chemical burn’, ‘acute poisoning’ or ‘asphyxiations’ (ESAW Injury codes 062, 071 and 081; not applicable for 2010). An additional 218 cases recorded as contact with hazardous substances via skin and eyes, inhalation and ingestion were included (ESAW contact—mode of injury codes 15–17, not applicable for 2010). Remaining 1139 cases were read case-by-case, searching for non-chemical incidents, for instance thermal burns and fall injuries, resulting in exclusion of 703 cases. After the hazard categorization, described in the next section, an additional 11 cases (5 biohazards and 6 thermal burns) were excluded resulting in 1237 cases in total. Figure 1. View largeDownload slide Schematic figure of ISA filtering procedure. Figure 1. View largeDownload slide Schematic figure of ISA filtering procedure. Severe accident/incident records SAIR reports were only available for 2012 through 2014. As the category ‘chemicals’ may not include all hazardous substances we also extracted cases reported as caused by ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’, yielding 10566 cases (Table 1, Fig. 2). Unlike ISA, SAIR contains reports concerning schoolchildren and workplace data were used to filter out schools in order to exclude cases concerning only schoolchildren. Like ISA, the entries under each causal category includes (potential) injuries by other means than harmful substances. Additional filters (Fig. 2) were applied to three of the five categories before we coded cases according to hazard and product group, which in turn assisted in excluding an additional number of cases as well as identifying multiple reports of the same injury/incident. The final SAIR excerpt for analysis contained 1208 cases, 391 injuries and 817 incidents. Figure 2. View largeDownload slide Schematic figure of SAIR filtering procedure, keywords are listed in the Supplementary material (available at Annals of Work Exposures and Health online). Figure 2. View largeDownload slide Schematic figure of SAIR filtering procedure, keywords are listed in the Supplementary material (available at Annals of Work Exposures and Health online). Poisons Information Centre For the years 2010 through 2014 the Swedish PIC has consulted on 8240 occupational cases involving adults (aged 20 or more, this cut-off is based on PIC age categories). As the ISA data and SAIR injury data concern confirmed injuries, we limited the PIC data to cases judged as major risk (n = 1736) and cases with confirmed severe injuries (n = 6) or fatalities (n = 1, included since this case concerns several individuals, of which one suffered fatal injuries), yielding 1743 cases. An overview of all occupational PIC cases is included in Supplementary Table S2 (available at Annals of Work Exposures and Health online). Categorization of chemical hazards and product groups Information about chemical agents is logged differently in the three databases (Table 1). For ISA, data information stated under ‘material agent’ was used for coding of chemical hazard group and product category. Not all reports mentioned a product or chemical agent under material agent, and many were unspecific (‘chemical’ or ‘gas’) or referred only to a source of the exposure agent (e.g. ‘canister’ or ‘pipe’). SAIR also does not systematically log information about product or chemical agent, but the information is generally retrievable from the free-text description of the event. The PIC recording of chemical agents often consists of two parts, the first referring to product group (e.g. pool chemical, cleaning agent) the second referring to chemical identity or hazard category (e.g. hypochlorite, alkali, irritant gas). However, as the priority of the PIC is to identify the hazardous agent and health risks, information on product group was frequently limited to ‘chemical’. More details on the chemical agent might also be specified in the free-text fields (collected for 129 cases). We coded the different data excerpts according to the same set of categories concerning nature of chemical hazard and product group. The categories were derived inductively through an iterative process of coding and recoding between the three excerpts. We used the PIC coding as a starting point, as this was the source most clearly and systematically identifying chemical hazards and product groups. In case of more than one category being relevant, the most severe acute hazard was used for categorization. Hazard/chemical group The categories of hazard or chemical groups used for the analyses are listed below with examples of included cases: Alkali: Sodium hydroxide, wet cement, lime. Acids: Hydrofluoric acid, battery acid, phosphoric acid Unspecified corrosives: Applicable to ISA only, exposures logged as causing chemical burns without further specifying chemical agent or hazard. Fumes, gases and vapours: Chlorine gas, fire fumes, exhaust fumes, ammonia. Hydrocarbons: Hydrocarbon solvents, hydraulic oil, fuels. Fibres and particles: Cement dust, stone dust, synthetic fibres, asbestos, powder from fire extinguisher. Other substances: Quaternary ammonium compounds, hypochlorite, pesticides, pharmaceuticals, metals and compounds, alcohols, glycols. Unspecified substances: E.g. chemical, gas, white powder, may also include reference to product group but not nature of chemical hazard, e.g. cleaning agent. Does not include ISA cases that caused corrosive injuries. For ISA, this also includes cases where the source of the exposure, e.g. hose or bucket, was specified rather than chemical agent. Product group Due to limitations in the available data, product groups are not defined at one particular level of supply chain use or sector. Only groups encompassing at least 3% of cases in one or more data excerpts were brought forward to the analyses, resulting in the following list: Bulk chemicals: Mainly covers substances or products used in bulk for manufacturing or processing, but may, in particular for PIC data, also indicate any kind of chemical product in any kind of occupational setting. Cleaning agents: Covers household and professional cleaning agents, dishwasher detergents, laundry detergents. Construction/demolition: E.g. wet cement, cement dust, sealing foams. Batteries: Leakage or explosions resulting in exposure to acid or alkaline electrolyte. Fire fumes: Due to fires, not including e.g. welding fumes or exhaust fumes. Disinfectants: E.g. surface- and hand disinfectants Lab chemical: Chemical agents used in laboratory work. Pharmaceutical: Pharmaceutical substances, for human or animal use. Car repair and care: E.g. glycols and rim cleaners, excluding car batteries. Not reported: Mainly applicable to ISA, e.g. when no specification of product category is recorded e.g. chemical, dust, gas or bucket. Data analysis It is not possible to ascertain how representative the data from each source are for the Swedish labour market. Hence, data analysis was limited to cross-tabulation and descriptive statistics. Results To estimate potential overlap between the 1237 ISA cases, 391 SAIR injuries, 817 SAIR incidents and 1743 PIC major risk cases, date of injury/incident were cross-tabulated and the results compiled into Fig. 3. For ISA and SAIR, we used the registered date of accident, while for PIC we used the date of first call for the case, assuming that most calls were made on the day of accident. As in many cases there is more than one accident registered for one particular day, this approach should yield an overestimate the overlap between databases. All of the databases cover a substantial share of unique dates, that is dates on which no accident/incident was recorded in any other database (Fig. 3). As severe injuries fall under both the requirements of the Social Insurance Code (ISA) and the Work Environment Act (SAIR), we expect SAIR injuries to be found among the ISA cases. Yet only 47 individual days had injuries registered in both ISA and SAIR, and of these only 20 cases in each excerpt were reported by the same workplace on the same date (data not shown). Hence, SAIR injury data presented herein largely represents an independent selection of cases rather than a subset of ISA. The largest relative overlap is instead found between ISA and PIC (n = 514), nevertheless only amounting to about half of the days covered by PIC. The overlap between PIC and SAIR injuries (n = 50) or SAIR incidents (n = 99) is even lower (Fig. 3). Figure 3. View largeDownload slide Venn diagram of overlap in date of injury/incident between the different databases. Figure 3. View largeDownload slide Venn diagram of overlap in date of injury/incident between the different databases. An increasing trend is seen in number of ISA cases per year, more specifically from 2011 to 2012, coinciding with the new online reporting form and changed benefit scheme, and from 2013 to 2014 (Table 2). The SAIR records, available for 2012 through 2014, also increase per year, particularly for incidents. The PIC major risk cases have on the other hand been relatively stable at approximately 350 major risk cases per year (although all occupational cases have increased, Supplementary Table S2, available at Annals of Work Exposures and Health online). Gender proportions are relatively similar between ISA cases and PIC major risk cases, among ISA cases, 29% of the injured were female, and the corresponding number for PIC major risk cases was 25% excluding cases with undetermined gender, SAIR data does not include gender (Table 2). Table 2. Number of cases per data excerpt for the years 2010 through 2014, gender, hazard category and product group (only groups >3% in at least one excerpt), in absolute numbers and relative frequency in % of respective data excerpt. ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 Percentages were listed in italics to contrast them to absolute numbers. Em dashes represent data unavailable or category not applicable. aLost days. bNear misses. cNinety-four percent of SAIR incidents concern asbestos. dNinety-seven percent of SAIR injuries and 92% of SAIR incidents, concern needle stick injuries where pharmaceutical (insulin, vaccine, anaesthetic) exposure was reported in parallel to concern for blood-borne diseases. View Large Table 2. Number of cases per data excerpt for the years 2010 through 2014, gender, hazard category and product group (only groups >3% in at least one excerpt), in absolute numbers and relative frequency in % of respective data excerpt. ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 Percentages were listed in italics to contrast them to absolute numbers. Em dashes represent data unavailable or category not applicable. aLost days. bNear misses. cNinety-four percent of SAIR incidents concern asbestos. dNinety-seven percent of SAIR injuries and 92% of SAIR incidents, concern needle stick injuries where pharmaceutical (insulin, vaccine, anaesthetic) exposure was reported in parallel to concern for blood-borne diseases. View Large Overall, corrosives were the most frequent hazards in ISA, SAIR injuries and PIC data (Table 2). Although almost two-thirds of ISA cases lacked any description of substance identity (unspecified corrosives and unspecified chemicals), the available injury information confirmed that corrosive hazards were most common. The interpretation of SAIR data, in particular for incidents, needs to consider the targeted manner in which cases were extracted for the present study. Under fumes, gases and vapours 41% of SAIR injuries (n = 33) and 46% of SAIR incidents (n = 83) concerned exposures to fire fumes (reported as cause ‘fire’). For SAIR incidents, the concern could be for other damages caused by a fire rather than health effects of fire fumes. In the case of fibres and particles, SAIR incident cases mostly concerned asbestos (n = 189, 131 reported as cause ‘other’), a substance whose long-term effects and strict regulations are generally well known in Sweden. As with the grouping of chemical hazards, a large share (48.3%) of ISA cases had no product group recorded (Table 2). Among identifiable product groups bulk chemicals, that is mainly substances or products used for manufacturing or processing, was the most frequently implicated group (22.0%). This group is followed by cleaning agents (10.3%); no other group reached the cut-off of 3% in the ISA database. Also among SAIR injuries, bulk chemicals were the most frequent group, followed by cleaning agents. In addition, construction/demolition, fire fumes, pharmaceuticals, batteries, lab chemicals fulfilled the 3% criterion. Among SAIR incidents, construction and demolition was most frequently implicated, under which almost all reports concerned asbestos, followed by the product group bulk chemicals. Pharmaceuticals (12.4%) and fire fumes (10.2%) were common among SAIR incidents, while at 4.7%, cleaning agents was only the fifth most common category. Pharmaceuticals found in SAIR generally concerned needle stick injuries (reported as cause ‘hand held-tools’) and concern for transmittable disease rather than the pharmaceutical residue could have been the main concern. In contrast, the two ISA cases, and the six PIC major risk cases involved injuries or concerns caused by the pharmaceutical (e.g. agents for animal euthanasia). Among PIC major risk cases, cleaning agents were the most frequently implicated product group (34.4%), followed by bulk chemicals (28.6%). Furthermore, construction/demolition, car repair and care products, batteries and disinfectants fulfilled the 3% criterion. Contrary to SAIR records, PIC major risk cases under the category construction/demolition did not include any asbestos exposures, as these generally do not cause acute health effects, but rather corrosive or irritant building materials such as wet cement. Discussion We have investigated and compared three Swedish sources of data on occupational injuries and incidents involving hazardous substances. In addition to provide knowledge that risk managers can take action on, an ideal database for investigating chemical accidents/incidents at work should be: (i) Complete, that is include all the chemical injuries and incidents that have taken place. (ii) Specific, that is allow chemical accidents/incidents to be distinguishable from other causes. (iii) Accurate with respect to reported exposure agents and injuries. The three databases investigated in the present study have different purposes, resulting in differences regarding these three criteria. In the following sections, we will address these criteria, discuss the findings of the comparisons between ISA, SAIR and PIC and then discuss potential and limitations of PIC data regarding occupational injury or hazard surveillance. Completeness of the three databases is affected by their different objectives and different incentives for those reporting. The reporting to the ISA system is designed for social insurance purposes, including workers’ compensation, and targets occupational injuries involving lost working days. Its unit of recording is injured individual, as opposed to SAIR and PIC that focus on accident/incident. However, while the connection to workers’ compensation likely is the main driver for reporting to ISA, it is noteworthy that it is possible for an individual to apply for workers compensation without prior reporting. The SAIR system covers severe injuries and incidents, but since ‘severe’ is poorly defined, the actual coverage is unclear. In addition, SAIR is without any economic incentive for employers and employees. Rather, the use of SAIR for inspection planning may deter employers from reporting, as an inspection could discover additional workplace issues needing attention and resources. Failing to report to ISA and SAIR as an employer is punishable by fines, although whether this constitutes a reporting incentive is questionable as there are few means to identify un-notified injuries or incidents. The PIC on the other hand advices about treatment and does not collect personal data or workplace data, hence calling would not risk incriminating the caller. Cases are recorded as occupational by the PIC expert if the caller identifies it as so. For the present study, we also assumed that PIC major risk cases are more comparable to ISA data than minor or moderate risk cases. However, there is no systematic investigation of the extent to which occupational PIC cases involve absence from work. Although PIC major risk cases are judged to need immediate healthcare, they will not necessarily result in lost days. Conversely, some moderate risk cases may require absence from work. As calling the PIC is voluntary, it is not appropriate to talk about completeness of PIC records. Nevertheless, the different objectives and incentives of these three systems mean that none of them covers all occupational accidents/incidents in Sweden. There are few investigations of the completeness of ISA; and, to our knowledge, none regarding completeness of SAIR. One resource for SWEA is the work-related disorders survey, distributed biannually to a sample of the population of working age (SWEA, 2014). Among respondents for 2014, 1.5% of women and 2.1% of men reported to have lost days due to a workplace accident in the preceding 12 months, corresponding to approximately 33000 women and 49000 men (SWEA, 2014). For all attributed causes, ISA covers 13471 cases for women and 17858 for men (http://webbstat.av.se/; last accessed 27 September 2017), indicating that three out of five lost day injuries were unreported in 2014. Systematic comparison between the ISA data and the work-related disorders survey was last performed for the years 1995–2000 (SCB, 2003). This comparison stated to have found 55% of the self-reported lost days injuries in ISA. While the work-related disorders survey offers the most robust estimate of the total burden of chemical injuries across occupations, it does not allow detailed analysis of chemical exposures as only one predefined category is available for causes connected to hazardous substances (‘chemical and technical product, mineral substances’, authors’ translation). In the 2014 survey, 0.6% of respondents (corresponding to more than 30 000 individuals in the Swedish workforce) reported to have suffered from any kind of work related disorder due to chemicals. The SWEA report does not distinguish between disorders caused by chemicals due to accidents or due to other workplace conditions, likely a majority of these disorders were due to other workplace conditions (SWEA, 2014). The increasing number of injuries found in ISA and SAIR over time could indicate that their coverage has improved. For 2010 through 2014 the increase in ISA was 71%. During the same period, the number of employed persons increased by 5% (Swedish workforce statistics from: http://www.statistikdatabasen.scb.se, accessed 27 September 2017). A large increase (72%) was also seen for SAIR incidents from 2012 to 2014, and a smaller for SAIR injuries (19%). While it is possible that the ISA and SAIR numbers reflect an actual increase in injuries/incidents, at least two factors suggest an increased reporting frequency. First, the change in the benefit scheme starting in April 2012, which may have increased propensity to report to ISA as well as likelihood for absence from work (SWEA, 2013). Second, the introduction of the online reporting interface for both ISA and SAIR (anmalaenarbetsskada.se, December 2011) facilitating employers’ reporting. In particular, the increasing numbers for SAIR incidents highlight the importance of simplifying employers’ reporting. That the SAIR injuries did not increase as notably may indicate that severe injuries are decreasing thus offsetting an increased reporting frequency. It is also conceivable that the reporting frequency is not increasing as markedly for this category because employers perceive the ISA reporting as sufficient (and/or the most suitable option of the available reporting forms on the portal page). No increase was seen for PIC major risk cases, although all occupational cases increased by 30% (Schenk et al., 2018). One potential explanation for this overall increase could be the precautionary statements introduced on labelling by the CLP regulation, of which some refer to PICs. However, there are also indications of a general increased awareness of PIC as also non-occupational cases have increased during the investigated period (Schenk et al., 2018). The different purposes of ISA, SAIR and PIC also influence the information collected about the injuries and incidents. Both ISA and SAIR focus on underlying causes, for example pipe-failure, rather than the nature of the chemical hazard, for example sodium hydroxide. This may allow identification of more general preventive measures, but will not allow identification for instance of products with problematic uses. PIC experts, on the other hand, have knowledge and resources at hand to identify the major hazard of the exposure. Hence, the PIC data provide reliable data on the identification of chemicals and products, in addition to route of exposure and a medical severity scoring. The focus of ISA hampers simple identification of injuries involving hazardous substances. Our starting point was cases registered as caused by ‘dusts, gases, fumes or liquids’ (n = 1951). However, after scrutiny of the data for these cases we concluded that at least 37% of these did not involve a chemical injury. Our targeted SAIR excerpt identified 391 injuries and 817 incidents, although future analyses of chemical injuries/incidents using SAIR data could be restricted to only injuries/incidents caused by chemicals as the high proportion of certain kinds of hazardous substances found under other categories (e.g. fire fumes and pharmaceuticals) may reflect concern for other hazards than toxicological. Nevertheless, a comparison of the dates and workplaces recorded in ISA and SAIR injuries shows only 20 cases overlap for 2012 through 2014, which is low also if only taking the SAIR category of chemicals into account (n = 298). As SAIR injuries should be a subset of ISA, this shows that ISA is not specific with regards to identifying cases involving hazardous substances. Possibly employers reporting to ISA perceive certain exposures as more suitable under other categories (e.g. ‘electricity, explosion or fire’ for fire fumes or exploding batteries). Hence, other sorting criteria, based on injury or contact descriptors, as included in the EU level ESAW methodology, may be more relevant for future reviews. Still, some difficulties with identifying chemical accidents will likely remain, as reporting of occupational injuries is highly dependent on coder even when guided by an ESAW compatible reporting form (Molinero-Ruiz et al., 2015; Jacinto et al., 2016). Furthermore, to identify of the nature of chemical hazard in ISA required a non-trivial coding effort by combining information under material agent and injury. For one-fourth of the ISA cases corrosion hazards were identified due to the recorded injuries, for another 41% hazards were not identifiable at all. Similarly, we were unable to connect 44% of ISA cases to any product group. For SAIR, we used the free-text description to identify chemical hazard (possible in 95% of cases), also a time-consuming procedure unfeasible for frequently repeated surveillance efforts. Hence, from the perspective of investigating acute chemical exposures at work, additional information sources with readily available hazard or product information are valuable. The lack of specificity regarding the identity of hazardous substances is connected to the accuracy of exposure data. Both ISA and SAIR are based on employers reporting, and as discussed above how they code accidents/incidents in the notification forms may differ significantly between individuals. Moreover, there may be incentives to not report circumstances accurately or in detail. PIC records also depend on the callers’ account, but trained experts enter the records and ask for the details needed for giving an appropriate advice, a target that also is in the interest of the caller. The lacking specificity and accuracy could partly be addressed by SWEA by adding more guidance on how to report material agents to the online reporting form. Addition of a reporting item with predefined exposure categories for cases involving hazardous substances to the ISA and SAIR reporting schemes could further enhance the specificity and accuracy of these data. The analysis of overlapping dates showed that the three databases largely cover different accidents/incidents. The comparison of hazards and products groups also shows differences between ISA, SAIR and PIC. Cleaning agents was the most common product group among PIC major risk cases and although relatively common also in ISA and SAIR, they were notably less so. This could indicate a systematic under-reporting of incidents with cleaning agents to SWEA. The lower frequency among SAIR incidents implies that cleaning agents are not perceived as having the potential for severe injuries. The hazards identified in our material were generally corrosion or irritation, due to alkali or acid properties. Acids and alkali in cleaning agents have also been frequently implicated for chemical ocular injuries (Tschopp et al., 2015). Suleiman and Svendsen (2014) found that safety data sheets for cleaning agents were inaccurate and/or incomplete, not allowing identification of all hazards or sufficient risk management measures. Furthermore, training of cleaning workers was found ineffective from a chemicals risk management perspective, as providers of training were lacking competence and as chemical health hazards were given comparably low priority (Suleiman and Svendsen 2015). Cleaning agents are used in many sectors, in some of which precarious employments are commonplace, which may decrease propensity to report injuries (Probst et al., 2013; Picchio and van Ours, 2017). Further research on the issue of acute exposures to cleaning agents, the causes and consequences is thus warranted. Turning to the use of PIC data in occupational injury and hazard surveillance, we find that the data offer complementary knowledge but that there are several factors limiting the use of PIC data for surveillance purposes. For instance, while in the present work cleaning agents were identified as an under-recognized hazard, PIC data offer no guidance on which kind of workplaces these hazards occur. Hence, PIC data alone cannot guide SWEA’s workplace inspections, or other kinds of workplace interventions. Furthermore, although a follow-up analysis of ISA and/or SAIR data could provide workplace data, the PIC data may concern a larger range of workplaces than covered by ISA. Nevertheless, by identifying product groups interventions could be targeted to relevant suppliers, who do inform downstream users about proper uses and risk management measures through the safety data sheets. For such identification of injury hazards also the PIC minor and moderate risk cases could be valuable to analyse as they, unlike ISA and SAIR accidents, offer opportunities for experience feed-back from incidents with no to minor injuries (Schenk et al., 2018). Although this is also the case for SAIR incidents, the PIC covers a significantly larger amount of cases; in 2014 the PIC handled 1831 occupational cases, five times as many as we identified for hazardous substances in SAIR incidents (Supplementary Table S2, available at Annals of Work Exposures and Health online). However, for continuous occupational injury surveillance regarding injuries caused by hazardous substances it might be more appropriate to improve the accuracy and specificity of ISA and SAIR reporting. Summary and conclusions None of the three database investigated in the present study can be expected to be complete, that is cover all accidents/incidents taking place at Swedish workplaces. We identified issues regarding specificity for ISA, for which it was challenging to separate out the cases involving hazardous substances. Furthermore, neither ISA not SAIR systematically include data on the identity of the hazardous substance, restricting identification of hazards and chemical products. Accuracy of the recorded data is an issue for all three databases. Both ISA and SAIR depend on employers’ reporting, and hence employers’ interpretation of the notification form as well as their truthfulness and attention to detail. The PIC data are recorded by trained experts who ask callers for the information needed to assess hazards and risks. Due to the nature of the PIC service, callers likely have a larger incentive to be detailed and truthful about involved hazardous substances. The strengths of ISA and SAIR are the recording of demographic and/or workplace data, allowing specific follow-up, and the identification of causes of exposures, such as machine failure, which may help identifying general issues behind accidents with a broad range of hazardous substances. The major strength of the PIC data is the identification of hazard and product group. Our comparisons of injury/incident dates between the different databases indicate significant differences in coverage of the different datasets, hence combining knowledge from all three sources should yield a more comprehensive picture of accidents/incidents at Swedish workplaces. In the present study, the comparison of chemical hazards and product groups represented in the different datasets identified cleaning agents as an under-recognized workplace hazard by SWEA data. Given the lack of specificity of SWEA data, periodical reviews of occupational cases handled by the Swedish PIC provide complementary knowledge regarding acute exposures to hazardous substances and may identify needs for further research and/or additional risk management efforts. Occupational injury reporting differs between countries and regions (Hämäläinen et al., 2006). Hence, transfer of our findings to other countries need to take national injury reporting systems and PIC coverage into account. Nevertheless, we find that co-operation between PICs and work environment authorities would improve the knowledge basis on acute chemical exposures at work, which in turn could be used to improve of the working environment. Supplementary Data Supplementary data are available at Annals of Work Exposures and Health online. Funding Funding for this research was provided by AFA Insurance (dnr 130288), an organization owned by Sweden’s labour market parties. The research at Swetox was supported by the Stockholm County Council, the Knut & Alice Wallenberg Foundation, and the Swedish Research Council FORMAS. Acknowledgements The authors thank Karin Feychting and Anita Annas, The Swedish Poisons Information centre, and Kjell Blom, the Swedish Work Environment Authority, for assistance with data retrieval and informative discussions about the respective data sources. Two anonymous reviewers are gratefully acknowledged for insightful comments and suggestions. The authors declare no conflict of interest relating to the material presented in this article. Its contents, including any opinions and/or conclusions expressed, are solely those of the authors. References Blanc PD , Olson KR . ( 1986 ) Occupationally related illness reported to a regional poison control center . Am J Public Health ; 76 : 1303 – 7 . Google Scholar CrossRef Search ADS PubMed Eurostat . ( 2013 ) European Statistics on Accidents at Work (ESAW) summary methodology . Available at http://ec.europa.eu/eurostat/documents/3859598/5926181/KS-RA-12-102-EN.PDF/56cd35ba-1e8a-4af3-9f9a-b3c47611ff1c. Accessed 27 September 2017 . Fagan KM , Hodgson MJ . ( 2017 ) Under-recording of work-related injuries and illnesses: an OSHA priority . J Safety Res ; 60 : 79 – 83 . Google Scholar CrossRef Search ADS PubMed Gravseth HM , Wegeland E , Lund J . ( 2003 ) Underrapportering av arbeidsskader til Arbeidstilsynet [Underreporting of occupational injuries to the labour inspectorate] . Tidskr Nor Lægefor ; 15 : 2057 – 59 . In Norwegian, English abstract available. Hämäläinen P , Takala J , Saarela KL . ( 2006 ) Global estimates of occupational accidents . Saf Sci ; 44 : 137 – 56 . Google Scholar CrossRef Search ADS Hinnen U , Hotz P , Gossweiler B et al. ( 1994 ) Surveillance of occupational illness through a national poison control center: an approach to reach small-scale enterprises ? Int Arch Occup Environ Health ; 66 : 117 – 23 . Google Scholar CrossRef Search ADS PubMed Jacinto C , Aspinwall E . ( 2004 ) A survey on occupational accidents’ reporting and registration systems in the European Union . Saf Sci ; 42 : 933 – 60 . Google Scholar CrossRef Search ADS Jacinto C , Santos FP , Guedes Soares C et al. ( 2016 ) Assessing the coding reliability of work accidents statistical data: how coders make a difference . J Safety Res ; 59 : 9 – 21 . Google Scholar CrossRef Search ADS PubMed Leigh JP , Marcin JP , Miller TR . ( 2004 ) An estimate of the U.S. Government’s undercount of nonfatal occupational injuries . J Occup Environ Med ; 46 : 10 – 8 . Google Scholar CrossRef Search ADS PubMed Litovitz T , Oderda G , White JD et al. ( 1993 ) Occupational and environmental exposures reported to poison centers . Am J Public Health ; 83 : 739 – 43 . Google Scholar CrossRef Search ADS PubMed Meulenbelt J , de Vries I . ( 1997 ) Acute work-related poisoning by pesticides in The Netherlands; a one year follow-up study . Przegl Lek ; 54 : 665 – 70 . Google Scholar PubMed Molinero-Ruiz E , Pitarque S , Fondevila-McDonald Y et al. ( 2015 ) How reliable and valid is the coding of the variables of the European Statistics on Accidents at Work (ESAW)? A need to improve preventive public policies . Saf Sci ; 79 : 72 – 9 . Google Scholar CrossRef Search ADS Olson DK , Sax L , Gunderson P et al. ( 1991 ) Pesticide poisoning surveillance through regional poison control centers . Am J Public Health ; 81 : 750 – 3 . Google Scholar CrossRef Search ADS PubMed Persson HE , Sjöberg GK , Haines JA et al. ( 1998 ) Poisoning severity score. Grading of acute poisoning . J Toxicol Clin Toxicol ; 36 : 205 – 13 . Google Scholar CrossRef Search ADS PubMed Picchio M , van Ours JC . ( 2017 ) Temporary jobs and the severity of workplace accidents . J Safety Res ; 61 : 41 – 51 . Google Scholar CrossRef Search ADS PubMed Probst TM , Barbaranelli C , Petitta L . ( 2013 ) The relationship between job insecurity and accident under-reporting: a test in two countries . Work Stress ; 27 : 383 – 402 . Google Scholar CrossRef Search ADS Rappin CL , Wuellner SE , Bonauto DK . ( 2016 ) Employer reasons for failing to report eligible workers’ compensation claims in the BLS survey of occupational injuries and illnesses . Am J Ind Med ; 59 : 343 – 56 . Google Scholar CrossRef Search ADS PubMed Rosenman KD , Kalush A , Reilly MJ et al. ( 2006 ) How much work-related injury and illness is missed by the current national surveillance system ? J Occup Environ Med ; 48 : 357 – 65 . Google Scholar CrossRef Search ADS PubMed Rubenstein H , Bresnitz EA . ( 2001 ) The utility of Poison Control Center data for assessing toxic occupational exposures among young workers . J Occup Environ Med ; 43 : 463 – 6 . Google Scholar CrossRef Search ADS PubMed SCB . ( 2003 ) Informationssystemet om arbetsskador och undersökningen om arbetsorsakade besvär. En jämförande studie. [Information system on occupational injuries and the work-related disorders survey. A comparative study] Report 2003 :5. p. 44 . Available at http://www.scb.se/statistik/AM/AM0602/1995I00/AM0602_1995I00_AM76ST0305.pdf. Accessed 27 September 2017 . Schenk L , Feychting K , Annas K et al. ( 2018 ) Records from the Swedish Poisons Information Centre as a means for surveillance of occupational accidents and incidents with chemicals . Saf Sci; 104: 269–75 . Sudakin DL , Power LE . ( 2007 ) Pyrethrin and pyrethroid exposures in the United States: a longitudinal analysis of incidents reported to poison centers . J Med Toxicol ; 3 : 94 – 9 . Google Scholar CrossRef Search ADS PubMed Suleiman AM , Svendsen KV . ( 2014 ) Are safety data sheets for cleaning products used in Norway a factor contributing to the risk of workers exposure to chemicals ? Int J Occup Med Environ Health ; 27 : 840 – 53 . Google Scholar CrossRef Search ADS PubMed Suleiman AM , Svendsen KV . ( 2015 ) Effectuality of cleaning workers’ training and cleaning enterprises’ chemical health hazard risk profiling . Saf Health Work ; 6 : 345 – 52 . Google Scholar CrossRef Search ADS PubMed Swedish Work Environment Authority (SWEA) . ( 2013 ) Arbetsskador 2012. Workrelated injuries 2012 . p. 102 . In Swedish. Available at https://www.av.se/globalassets/filer/statistik/arbetsmiljostatistik-arbetsskador-2012-rapport-2013-01.pdf. Accessed 27 September 2017 . Swedish Work Environment Authority (SWEA) . ( 2014 ) Work-related disorders 2014. Report 2014 :4. Available at https://www.av.se/globalassets/filer/statistik/arbetsmiljostatistik-arbetsorsakade-besvar-rapport-2014.pdf. Accessed 27 September 2017 . Swedish Work Environment Authority (SWEA) . ( 2015 ) Årsredovisning 2014. Yearly report 2014 . p. 52 . In Swedish. Available at https://www.av.se/globalassets/filer/om-oss/arsredovisningar/arsredovisning-arbetsmiljoverket-2014.pdf. Accessed 27 September 2017 . Trueblood AB , Shipp EM . ( 2017 ) Characteristics of occupational pesticide exposures reported to poinsin control centers in Texas, 2000–2015 . Arch Environ Occup Health ; 12 : 1 – 8 . Google Scholar CrossRef Search ADS Tschopp M , Krähenbühl P , Tappeiner C et al. ( 2015 ) Incidence and causative agents of chemical eye injuries in Switzerland . Clin Toxicol (Phila) ; 53 : 957 – 61 . Google Scholar CrossRef Search ADS PubMed United States Government Accountability Office (US GAO) . ( 2012 ) Workplace safety and health: better OSHA guidance needed on safety incentive programs . GAO-12– 329. Available at http://www.gao.gov/products/GAO- 12–329. Accessed 27 September 2017 . Woolf A , Alpert HR , Garg A et al. ( 2001 ) Adolescent occupational toxic exposures: a national study . Arch Pediatr Adolesc Med ; 155 : 704 – 10 . Google Scholar CrossRef Search ADS PubMed Ziqubu-Page T , Forrester MB . ( 2016 ) Adolescent workplace exposures reported to Texan poison centers . Int J Adolesc Med Health . [Epub ahead of print] .doi: 10.1515/ijamh-2016-0057 © The Author(s) 2018. Published by Oxford University Press on behalf of the British Occupational Hygiene Society. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Annals of Work Exposures and Health (formerly Annals Of Occupational Hygiene) Oxford University Press

Comparing Data from the Poisons Information Centre with Employers’ Accident Reports Reveal Under-Recognized Hazards at the Workplace

Loading next page...
 
/lp/ou_press/comparing-data-from-the-poisons-information-centre-with-employers-Ftg4UnR9Ew
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the British Occupational Hygiene Society.
ISSN
2398-7308
eISSN
2398-7316
D.O.I.
10.1093/annweh/wxy009
Publisher site
See Article on Publisher Site

Abstract

Abstract Records of injuries and incidents provide an important basis for injury prevention related to hazardous substances at the workplace. The present study aimed to review available data on injuries and incidents involving hazardous substances and investigate how data from the Poisons Information Centre could complement the records of the Swedish Work Environment Authority. We found two major obstacles for using injury/incident data based on employers’ mandatory reporting. First, it was not possible to quickly and reliably identify injuries caused by hazardous substances, and second, data identifying substances or products are not systematically included. For two out of five investigated injuries with lost working days likely due to chemical injuries, we could not identify substances and/or products involved. The records based on calls to the Poisons Information Centre allow better understanding of chemical hazards and products. Besides the large share of unidentified chemical hazards in the injury statistics, the most striking difference was found for cleaning agents. Cleaning agents were implicated in one-third of the occupational cases that the consulting Poisons Information Centre expert judged to pose a major risk and in need of immediate healthcare. Only one in 10 injuries with lost days reported by employers was related to this type of product. The identification of exposures and symptoms by the Poisons Information Centre allow recognition of chemicals with problematic occupational uses. Hence, these records may serve as an important complement to official injury statistics related to incidents with hazardous substances at work. chemicals management, occupational health and safety, poison control, risk assessment, work injury Introduction Hazardous substances in the workplace cause a wide variety of occupational injuries and diseases. A part of the authority risk management regime is to compile information on occupational diseases, injuries and severe incidents. Analysis of this information makes up part of the authorities’ work to identify areas of interest for inspection or information efforts (Jacinto and Aspinwall, 2004). However, we have reasons to believe that occupational injury and disease statistics are incomplete and underestimate the number of injuries. Certain groups of workers may be undercounted because they fall outside the reporting requirements, such as trainees, self-employed or from the informal sector. Studies from the United States have estimated that up to two-thirds of occupational injuries are not included in the national surveillance system (Leigh et al., 2004; Rosenman et al., 2006). Under-reporting is also a cause of incomplete injury statistics. Employers may lack awareness of the legal requirements to report injuries and incidents, reporting responsibilities may be confused due to the (increasingly common) complex employment structures in many industrial sectors (Rappin et al., 2016). It is also plausible that incentives are low to report and incriminate oneself or to make the effort in the face of limited time. Furthermore, under-reporting may be due to employees’ under-reporting to their employers, for instance due to disciplinary programs or incentive programs that penalize accidents (US GAO, 2012; Fagan and Hodgson, 2017). Moreover, any underestimate is likely skewed. For instance, the degree of under-reporting has been found to be larger for less severe injuries (Gravseth et al., 2003). In addition, job insecurity has been found to correlate with both an increased risk for injury and a decreased reporting of injury (Probst et al., 2013; Picchio and van Ours, 2017). Undercounting and under-reporting of the number of chemical incidents and injuries reduces the ability to identify and address occupational safety issues with chemicals. One implication of skewed underestimates is a skewed decision-basis for the authorities, which may further reduce efficiency of their risk management measures. Additional data sources, which cover a larger or a different subset of chemical injuries and incidents, may thus prove to be valuable complements to official injury data. Data from Poisons Information Centres (PIC) have been used to provide overviews of accidents with hazardous substances at work (Blanc and Olson, 1986; Litovitz et al., 1993; Schenk et al., 2018). PICs inform the public, workplaces and/or healthcare personnel regarding the toxicity of exposure agents and the management of exposed patients. In performing these tasks, the PICs gather data regarding exposures that may be useful for managing the incident or for surveillance and follow-up. Although not a standard tool for occupational injury surveillance, such data have been used to investigate a variety of occupational issues or groups, such as exposure to pesticides (Olson et al., 1991; Meulenbelt and de Vries, 1997; Sudakin and Power, 2007; Trueblood and Shipp, 2017), workers in small enterprises (Hinnen et al., 1994) or adolescent workers (Rubenstein and Bresnitz, 2001; Woolf et al., 2001; Ziqubu-Page and Forrester, 2016). A recent overview of Swedish PIC data on occupational cases indicated that these might cover a larger range of occupational injuries or incidents involving hazardous substances than registered by the national occupational injury statistics (Schenk et al., 2018). The objective of the present work was to compare the injuries and incidents involving hazardous substances recorded under Swedish occupational injury data and in the call records of the national PIC. Specific questions addressed were: • Do the different databases cover a largely different set of injuries and incidents? • Are there differences in the patterns of chemical hazards or product groups encountered in these databases? • What are the strengths and limitations of the respective databases for investigating accidents/incidents with hazardous substances? Through addressing these questions we aimed to evaluate if and to what extent data from the PIC provide complementary knowledge that can be used for prevention, for instance by identifying hazards or chemical products likely to be under-recognized in the occupational injury data. Methods In the following sections, we describe how data were retrieved, managed and analysed. Ethical vetting was applied for and approved by the regional ethical board in Stockholm (2015/1-31/5). Ethical vetting was not sought for matching individuals case by case between data sources. Data sources Table 1 outlines the three data sources. Two of the databases are managed by the Swedish Work Environment Authority (SWEA), which provided the data excerpts used in the present study. The national statistics are compiled in a database named ISA (acronym for information system about occupational injuries). This database is based on employers’ reporting under the Swedish Social Insurance Code (Socialförsäkringsbalk 2010:110), which requires employers to report work-related injuries. If such injuries involve one or more days of absence from work, they are registered in ISA. Parts of the ISA database are publicly available online (http://webbstat.av.se/; last accessed 27 September 2017). Reporting to ISA follows a structured notification form including variables defined by SWEA and the European Statistics on Accidents at Work (ESAW; Eurostat, 2013). ESAW was under implementation during the period of the present study (Commission Directive 349/2011), when referring to ESAW variables the period of applicability will be specified. Since December 2011, ISA reports can be submitted online (https://anmalarbetsskada.se/, last accessed 27 September 2017). Also, the occupational injury benefit scheme was changed for a large part of the Swedish workforce in April 2012, with injury compensation starting from the first day of sickness absence, compared to previously from the second day of absence. This would provide an incentive for the employees to require documentation of occupational injuries under the Social Insurance Code (i.e. ISA) and possibly increase likelihood for absence from work after an injury (SWEA, 2013). Table 1. Overview of databases. ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 aAll injuries should be reported under the social insurance legislation, only >1 lost days are registered in ISA. bSevere injury is not defined, se text for examples given by SWEA of severe injuries. cAll categories at this level were ‘electricity’, ‘fire’, ‘explosion’, ‘chemical’, ‘falling or flying objects’, ‘fall of person’, ‘animals’, ‘vehicles’, ‘physical overload or misstep’, ‘machine or transport device’, hand-held tool or object’(includes syringes), ‘threat’, ‘robbery’, ‘violence’, ‘psychological chock’ and ‘other cause’. dAll categories at this level (accident was caused by) were ‘dusts, gases, fumes or liquids’, ‘electricity, explosion or fire’, ‘material fell, plummeted, burst or exploded’, ‘collision between vehicles, hit by vehicle, injured by object, machine, tool or animal’, ‘person fell’, ‘struck against something, stepped on sharp item’, ‘misstep, lift or other overload’, ‘threats or violence, chock, fear’ and ‘other cause’. These categories are unaffected by the implementation of the ESAW methodology. View Large Table 1. Overview of databases. ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 ISA SAIR PIC Owner of database SWEA SWEA PIC Recorded injuries/incidents ≥1 lost daysa (excluding fatalities) Severeb injuries or severe incidents All calls regardless of severity Reporting required under Social Insurance Legislation (Chap. 42 10§) Work Environment Legislation (Chap. 3 3a§) Not required Reporting agent Employer Employer Variable (public, healthcare actors, etc.) Possible consequence of non-compliance Fines Fines Not applicable Reporting/recording means Webpage (since December 2011) structured notification form Webpage (since December 2011), structured notification form Telephone, PIC expert fills in a structured recording form during consultation Purpose Records for injury compensation, basis for prioritization and follow-up of SWEA measures Plan inspections and experience feedback Offer advice on treatment Focus Cause of accident, injury and workplace Type of accident/incident and workplace Exposure and symptoms Case unit Injured individual Accident/incident Accident/incident Collected data  Demographic data of the employee Yes No (recorded for injuries, not available) Only age category and gender  Workplace data Yes Yes No (possibly in free-text)  Date and time Of injury and of report Of injury/incident and of report Of call  Cause of accident/incident Immediate cause, structured (SWEA specific and ESAW variables in several levels). Free-text description (not included). Yes, structured (16 SWEA specific categoriesc) and free-text description of event No (possibly in free-text)  Type and site of injury Yes (ESAW variables) No (possibly in free-text) No (possibly in free-text)  Exposure route Indirectly (ESAW variables for contact and site of injury) No (possibly in free-text) Yes  Chemical agent Extracted from ‘material agent’ (ESAW variable) Extracted from free-text description of event Yes (according to predefined categories)  Severity As days lost Injury or incidents; Number of injured Risk estimates or poisoning severity score  Free-text fields None collected Description of event Question and answer  Time period collected 2010 through 2014 2012 through 2014 2010 through 2014  First filtering criteria Cause: ‘dust, gases, fumes or liquids’d Cause: ‘chemicals’, ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’c Occupational incidents involving adults (aged 20 or more) judged to pose a major risk  Cases in initial excerpt 1951 10566 1743  Cases included 1237 1208 (391 injuries; 817 incidents) 1743 aAll injuries should be reported under the social insurance legislation, only >1 lost days are registered in ISA. bSevere injury is not defined, se text for examples given by SWEA of severe injuries. cAll categories at this level were ‘electricity’, ‘fire’, ‘explosion’, ‘chemical’, ‘falling or flying objects’, ‘fall of person’, ‘animals’, ‘vehicles’, ‘physical overload or misstep’, ‘machine or transport device’, hand-held tool or object’(includes syringes), ‘threat’, ‘robbery’, ‘violence’, ‘psychological chock’ and ‘other cause’. dAll categories at this level (accident was caused by) were ‘dusts, gases, fumes or liquids’, ‘electricity, explosion or fire’, ‘material fell, plummeted, burst or exploded’, ‘collision between vehicles, hit by vehicle, injured by object, machine, tool or animal’, ‘person fell’, ‘struck against something, stepped on sharp item’, ‘misstep, lift or other overload’, ‘threats or violence, chock, fear’ and ‘other cause’. These categories are unaffected by the implementation of the ESAW methodology. View Large In addition to ISA reporting, severe injuries and severe incidents are also to be reported separately to the SWEA under the Swedish Work Environment Act (Arbetsmiljölag 1977:1160). There is no strict definition of severe injury; injuries to internal organs, burns larger than 5% of the body surface, and accidents where several persons were injured are given as examples. Severe incidents are defined as near misses that could have caused severe injuries. These severe accident/incident records (henceforth, SAIR) are not publicly available, but are used by the SWEA in planning inspection efforts. In 2014, 13% of first time workplace inspections were in response to SAIR reports (SWEA, 2015). The SAIR notification form is shorter than that of ISA and uses different categories for cause (Table 1); electronic reporting was launched at the same date and through the same webpage as ISA. In Sweden, one PIC unit serves the whole country (population of 10 million). The 24-h phone service is open to medical professionals as well as the general public. Telephone consultations are logged in the PIC database and multiple calls about the same case are connected to the first call allowing identification of individual cases. The Swedish PIC is responsible for receiving information relating to composition of hazardous products under the EU Regulation (EC) 1272/2008 on classification, labelling and packaging (CLP). Hence, PIC experts have access to a database on product composition, aiding in the identification of major hazards when provided with the product name. Based on reported exposures (extent and agent) and symptoms, the PIC expert assesses the risk to the exposed. Minor risk cases present with no or minor symptoms and are manageable on site. Moderate risk cases may present pronounced and/or prolonged symptoms, immediate healthcare may be recommended by the PIC expert. Major risk cases present or may present severe symptoms and immediate healthcare is recommended. If the PIC expert is unable to evaluate risk, for example due to limited information, this is logged as undetermined risk. If applicable, risk estimates are replaced by confirmed severity corresponding to poisoning severity score (PSS) 3-severe injury or 4-fatality (Persson et al., 1998). Extraction of relevant cases Information system about occupational injuries The cause-of-injury category ‘dust, gases, fumes or liquids’ was used as the most relevant category for exposures to hazardous substances, yielding 1951 cases for 2010 through 2014 (Table 1). As these cases may include non-poisoning hazards, for example burns by boiling water, the records needed additional filtering (Fig. 1). Based on injury data, 594 cases were included, that is ‘chemical burn’, ‘acute poisoning’ or ‘asphyxiations’ (ESAW Injury codes 062, 071 and 081; not applicable for 2010). An additional 218 cases recorded as contact with hazardous substances via skin and eyes, inhalation and ingestion were included (ESAW contact—mode of injury codes 15–17, not applicable for 2010). Remaining 1139 cases were read case-by-case, searching for non-chemical incidents, for instance thermal burns and fall injuries, resulting in exclusion of 703 cases. After the hazard categorization, described in the next section, an additional 11 cases (5 biohazards and 6 thermal burns) were excluded resulting in 1237 cases in total. Figure 1. View largeDownload slide Schematic figure of ISA filtering procedure. Figure 1. View largeDownload slide Schematic figure of ISA filtering procedure. Severe accident/incident records SAIR reports were only available for 2012 through 2014. As the category ‘chemicals’ may not include all hazardous substances we also extracted cases reported as caused by ‘explosions’, ‘fire’, ‘hand-held tool or object’ and ‘other cause’, yielding 10566 cases (Table 1, Fig. 2). Unlike ISA, SAIR contains reports concerning schoolchildren and workplace data were used to filter out schools in order to exclude cases concerning only schoolchildren. Like ISA, the entries under each causal category includes (potential) injuries by other means than harmful substances. Additional filters (Fig. 2) were applied to three of the five categories before we coded cases according to hazard and product group, which in turn assisted in excluding an additional number of cases as well as identifying multiple reports of the same injury/incident. The final SAIR excerpt for analysis contained 1208 cases, 391 injuries and 817 incidents. Figure 2. View largeDownload slide Schematic figure of SAIR filtering procedure, keywords are listed in the Supplementary material (available at Annals of Work Exposures and Health online). Figure 2. View largeDownload slide Schematic figure of SAIR filtering procedure, keywords are listed in the Supplementary material (available at Annals of Work Exposures and Health online). Poisons Information Centre For the years 2010 through 2014 the Swedish PIC has consulted on 8240 occupational cases involving adults (aged 20 or more, this cut-off is based on PIC age categories). As the ISA data and SAIR injury data concern confirmed injuries, we limited the PIC data to cases judged as major risk (n = 1736) and cases with confirmed severe injuries (n = 6) or fatalities (n = 1, included since this case concerns several individuals, of which one suffered fatal injuries), yielding 1743 cases. An overview of all occupational PIC cases is included in Supplementary Table S2 (available at Annals of Work Exposures and Health online). Categorization of chemical hazards and product groups Information about chemical agents is logged differently in the three databases (Table 1). For ISA, data information stated under ‘material agent’ was used for coding of chemical hazard group and product category. Not all reports mentioned a product or chemical agent under material agent, and many were unspecific (‘chemical’ or ‘gas’) or referred only to a source of the exposure agent (e.g. ‘canister’ or ‘pipe’). SAIR also does not systematically log information about product or chemical agent, but the information is generally retrievable from the free-text description of the event. The PIC recording of chemical agents often consists of two parts, the first referring to product group (e.g. pool chemical, cleaning agent) the second referring to chemical identity or hazard category (e.g. hypochlorite, alkali, irritant gas). However, as the priority of the PIC is to identify the hazardous agent and health risks, information on product group was frequently limited to ‘chemical’. More details on the chemical agent might also be specified in the free-text fields (collected for 129 cases). We coded the different data excerpts according to the same set of categories concerning nature of chemical hazard and product group. The categories were derived inductively through an iterative process of coding and recoding between the three excerpts. We used the PIC coding as a starting point, as this was the source most clearly and systematically identifying chemical hazards and product groups. In case of more than one category being relevant, the most severe acute hazard was used for categorization. Hazard/chemical group The categories of hazard or chemical groups used for the analyses are listed below with examples of included cases: Alkali: Sodium hydroxide, wet cement, lime. Acids: Hydrofluoric acid, battery acid, phosphoric acid Unspecified corrosives: Applicable to ISA only, exposures logged as causing chemical burns without further specifying chemical agent or hazard. Fumes, gases and vapours: Chlorine gas, fire fumes, exhaust fumes, ammonia. Hydrocarbons: Hydrocarbon solvents, hydraulic oil, fuels. Fibres and particles: Cement dust, stone dust, synthetic fibres, asbestos, powder from fire extinguisher. Other substances: Quaternary ammonium compounds, hypochlorite, pesticides, pharmaceuticals, metals and compounds, alcohols, glycols. Unspecified substances: E.g. chemical, gas, white powder, may also include reference to product group but not nature of chemical hazard, e.g. cleaning agent. Does not include ISA cases that caused corrosive injuries. For ISA, this also includes cases where the source of the exposure, e.g. hose or bucket, was specified rather than chemical agent. Product group Due to limitations in the available data, product groups are not defined at one particular level of supply chain use or sector. Only groups encompassing at least 3% of cases in one or more data excerpts were brought forward to the analyses, resulting in the following list: Bulk chemicals: Mainly covers substances or products used in bulk for manufacturing or processing, but may, in particular for PIC data, also indicate any kind of chemical product in any kind of occupational setting. Cleaning agents: Covers household and professional cleaning agents, dishwasher detergents, laundry detergents. Construction/demolition: E.g. wet cement, cement dust, sealing foams. Batteries: Leakage or explosions resulting in exposure to acid or alkaline electrolyte. Fire fumes: Due to fires, not including e.g. welding fumes or exhaust fumes. Disinfectants: E.g. surface- and hand disinfectants Lab chemical: Chemical agents used in laboratory work. Pharmaceutical: Pharmaceutical substances, for human or animal use. Car repair and care: E.g. glycols and rim cleaners, excluding car batteries. Not reported: Mainly applicable to ISA, e.g. when no specification of product category is recorded e.g. chemical, dust, gas or bucket. Data analysis It is not possible to ascertain how representative the data from each source are for the Swedish labour market. Hence, data analysis was limited to cross-tabulation and descriptive statistics. Results To estimate potential overlap between the 1237 ISA cases, 391 SAIR injuries, 817 SAIR incidents and 1743 PIC major risk cases, date of injury/incident were cross-tabulated and the results compiled into Fig. 3. For ISA and SAIR, we used the registered date of accident, while for PIC we used the date of first call for the case, assuming that most calls were made on the day of accident. As in many cases there is more than one accident registered for one particular day, this approach should yield an overestimate the overlap between databases. All of the databases cover a substantial share of unique dates, that is dates on which no accident/incident was recorded in any other database (Fig. 3). As severe injuries fall under both the requirements of the Social Insurance Code (ISA) and the Work Environment Act (SAIR), we expect SAIR injuries to be found among the ISA cases. Yet only 47 individual days had injuries registered in both ISA and SAIR, and of these only 20 cases in each excerpt were reported by the same workplace on the same date (data not shown). Hence, SAIR injury data presented herein largely represents an independent selection of cases rather than a subset of ISA. The largest relative overlap is instead found between ISA and PIC (n = 514), nevertheless only amounting to about half of the days covered by PIC. The overlap between PIC and SAIR injuries (n = 50) or SAIR incidents (n = 99) is even lower (Fig. 3). Figure 3. View largeDownload slide Venn diagram of overlap in date of injury/incident between the different databases. Figure 3. View largeDownload slide Venn diagram of overlap in date of injury/incident between the different databases. An increasing trend is seen in number of ISA cases per year, more specifically from 2011 to 2012, coinciding with the new online reporting form and changed benefit scheme, and from 2013 to 2014 (Table 2). The SAIR records, available for 2012 through 2014, also increase per year, particularly for incidents. The PIC major risk cases have on the other hand been relatively stable at approximately 350 major risk cases per year (although all occupational cases have increased, Supplementary Table S2, available at Annals of Work Exposures and Health online). Gender proportions are relatively similar between ISA cases and PIC major risk cases, among ISA cases, 29% of the injured were female, and the corresponding number for PIC major risk cases was 25% excluding cases with undetermined gender, SAIR data does not include gender (Table 2). Table 2. Number of cases per data excerpt for the years 2010 through 2014, gender, hazard category and product group (only groups >3% in at least one excerpt), in absolute numbers and relative frequency in % of respective data excerpt. ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 Percentages were listed in italics to contrast them to absolute numbers. Em dashes represent data unavailable or category not applicable. aLost days. bNear misses. cNinety-four percent of SAIR incidents concern asbestos. dNinety-seven percent of SAIR injuries and 92% of SAIR incidents, concern needle stick injuries where pharmaceutical (insulin, vaccine, anaesthetic) exposure was reported in parallel to concern for blood-borne diseases. View Large Table 2. Number of cases per data excerpt for the years 2010 through 2014, gender, hazard category and product group (only groups >3% in at least one excerpt), in absolute numbers and relative frequency in % of respective data excerpt. ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 ISA SAIR injurya SAIR incidentb PIC major risk n = 1237 n = 391 n = 817 n = 1743 Year n % n % n % n %  2010 189 15.3 — — — — 333 19.1  2011 185 15.0 — — — — 373 21.4  2012 269 21.7 118 30.2 201 24.6 346 19.9  2013 267 21.6 134 34.3 272 33.3 357 20.5  2014 324 26.2 140 35.8 347 42.5 334 19.2 Gender  Women 358 28.9 — — — — 418 24.0  Men 879 71.1 — — — — 1284 73.7  Unknown — — — — — — 41 2.4 Hazard/chemical group  Alkali 98 7.9 90 23 56 6.9 766 43.9  Acids 73 5.9 48 12.3 56 6.9 340 19.5  Unspecified corrosives 312 25.2 — — — — — —  Fumes, gases and vapours 122 9.9 80 20.5 181 22.2 175 10.0  Hydrocarbons 43 3.5 25 6.4 45 5.5 102 5.9  Fibres and particles 57 4.6 19 4.9 197 24.1 3 0.2  Other substances 29 2.3 104 26.6 249 30.5 280 16.1  Unspecified substances 503 40.7 25 6.4 33 4.0 77 4.4 Product group  Bulk chemicals 272 22.0 134 34.3 181 22.2 499 28.6  Cleaning agents 127 10.3 49 12.5 38 4.7 599 34.4  Construction/demolition 28 2.3 39 10.0 202c 24.7c 80 4.6  Batteries 27 2.2 21 5.4 14 2.6 54 3.1  Fire fumes 14 1.1 33 8.4 83 10.2 7 0.4  Disinfectants 8 0.6 6 1.5 7 0.9 54 3.1  Lab chemicals 4 0.3 12 3.1 51 6.2 37 2.1  Pharmaceutical 2 0.2 28d 7.2d 101d 12.4d 6 0.3  Car repair and care 3 0.2 1 0.3 8 1.0 65 3.7  Not reported 597 48.3 4 1.0 3 0.4 0 0 Percentages were listed in italics to contrast them to absolute numbers. Em dashes represent data unavailable or category not applicable. aLost days. bNear misses. cNinety-four percent of SAIR incidents concern asbestos. dNinety-seven percent of SAIR injuries and 92% of SAIR incidents, concern needle stick injuries where pharmaceutical (insulin, vaccine, anaesthetic) exposure was reported in parallel to concern for blood-borne diseases. View Large Overall, corrosives were the most frequent hazards in ISA, SAIR injuries and PIC data (Table 2). Although almost two-thirds of ISA cases lacked any description of substance identity (unspecified corrosives and unspecified chemicals), the available injury information confirmed that corrosive hazards were most common. The interpretation of SAIR data, in particular for incidents, needs to consider the targeted manner in which cases were extracted for the present study. Under fumes, gases and vapours 41% of SAIR injuries (n = 33) and 46% of SAIR incidents (n = 83) concerned exposures to fire fumes (reported as cause ‘fire’). For SAIR incidents, the concern could be for other damages caused by a fire rather than health effects of fire fumes. In the case of fibres and particles, SAIR incident cases mostly concerned asbestos (n = 189, 131 reported as cause ‘other’), a substance whose long-term effects and strict regulations are generally well known in Sweden. As with the grouping of chemical hazards, a large share (48.3%) of ISA cases had no product group recorded (Table 2). Among identifiable product groups bulk chemicals, that is mainly substances or products used for manufacturing or processing, was the most frequently implicated group (22.0%). This group is followed by cleaning agents (10.3%); no other group reached the cut-off of 3% in the ISA database. Also among SAIR injuries, bulk chemicals were the most frequent group, followed by cleaning agents. In addition, construction/demolition, fire fumes, pharmaceuticals, batteries, lab chemicals fulfilled the 3% criterion. Among SAIR incidents, construction and demolition was most frequently implicated, under which almost all reports concerned asbestos, followed by the product group bulk chemicals. Pharmaceuticals (12.4%) and fire fumes (10.2%) were common among SAIR incidents, while at 4.7%, cleaning agents was only the fifth most common category. Pharmaceuticals found in SAIR generally concerned needle stick injuries (reported as cause ‘hand held-tools’) and concern for transmittable disease rather than the pharmaceutical residue could have been the main concern. In contrast, the two ISA cases, and the six PIC major risk cases involved injuries or concerns caused by the pharmaceutical (e.g. agents for animal euthanasia). Among PIC major risk cases, cleaning agents were the most frequently implicated product group (34.4%), followed by bulk chemicals (28.6%). Furthermore, construction/demolition, car repair and care products, batteries and disinfectants fulfilled the 3% criterion. Contrary to SAIR records, PIC major risk cases under the category construction/demolition did not include any asbestos exposures, as these generally do not cause acute health effects, but rather corrosive or irritant building materials such as wet cement. Discussion We have investigated and compared three Swedish sources of data on occupational injuries and incidents involving hazardous substances. In addition to provide knowledge that risk managers can take action on, an ideal database for investigating chemical accidents/incidents at work should be: (i) Complete, that is include all the chemical injuries and incidents that have taken place. (ii) Specific, that is allow chemical accidents/incidents to be distinguishable from other causes. (iii) Accurate with respect to reported exposure agents and injuries. The three databases investigated in the present study have different purposes, resulting in differences regarding these three criteria. In the following sections, we will address these criteria, discuss the findings of the comparisons between ISA, SAIR and PIC and then discuss potential and limitations of PIC data regarding occupational injury or hazard surveillance. Completeness of the three databases is affected by their different objectives and different incentives for those reporting. The reporting to the ISA system is designed for social insurance purposes, including workers’ compensation, and targets occupational injuries involving lost working days. Its unit of recording is injured individual, as opposed to SAIR and PIC that focus on accident/incident. However, while the connection to workers’ compensation likely is the main driver for reporting to ISA, it is noteworthy that it is possible for an individual to apply for workers compensation without prior reporting. The SAIR system covers severe injuries and incidents, but since ‘severe’ is poorly defined, the actual coverage is unclear. In addition, SAIR is without any economic incentive for employers and employees. Rather, the use of SAIR for inspection planning may deter employers from reporting, as an inspection could discover additional workplace issues needing attention and resources. Failing to report to ISA and SAIR as an employer is punishable by fines, although whether this constitutes a reporting incentive is questionable as there are few means to identify un-notified injuries or incidents. The PIC on the other hand advices about treatment and does not collect personal data or workplace data, hence calling would not risk incriminating the caller. Cases are recorded as occupational by the PIC expert if the caller identifies it as so. For the present study, we also assumed that PIC major risk cases are more comparable to ISA data than minor or moderate risk cases. However, there is no systematic investigation of the extent to which occupational PIC cases involve absence from work. Although PIC major risk cases are judged to need immediate healthcare, they will not necessarily result in lost days. Conversely, some moderate risk cases may require absence from work. As calling the PIC is voluntary, it is not appropriate to talk about completeness of PIC records. Nevertheless, the different objectives and incentives of these three systems mean that none of them covers all occupational accidents/incidents in Sweden. There are few investigations of the completeness of ISA; and, to our knowledge, none regarding completeness of SAIR. One resource for SWEA is the work-related disorders survey, distributed biannually to a sample of the population of working age (SWEA, 2014). Among respondents for 2014, 1.5% of women and 2.1% of men reported to have lost days due to a workplace accident in the preceding 12 months, corresponding to approximately 33000 women and 49000 men (SWEA, 2014). For all attributed causes, ISA covers 13471 cases for women and 17858 for men (http://webbstat.av.se/; last accessed 27 September 2017), indicating that three out of five lost day injuries were unreported in 2014. Systematic comparison between the ISA data and the work-related disorders survey was last performed for the years 1995–2000 (SCB, 2003). This comparison stated to have found 55% of the self-reported lost days injuries in ISA. While the work-related disorders survey offers the most robust estimate of the total burden of chemical injuries across occupations, it does not allow detailed analysis of chemical exposures as only one predefined category is available for causes connected to hazardous substances (‘chemical and technical product, mineral substances’, authors’ translation). In the 2014 survey, 0.6% of respondents (corresponding to more than 30 000 individuals in the Swedish workforce) reported to have suffered from any kind of work related disorder due to chemicals. The SWEA report does not distinguish between disorders caused by chemicals due to accidents or due to other workplace conditions, likely a majority of these disorders were due to other workplace conditions (SWEA, 2014). The increasing number of injuries found in ISA and SAIR over time could indicate that their coverage has improved. For 2010 through 2014 the increase in ISA was 71%. During the same period, the number of employed persons increased by 5% (Swedish workforce statistics from: http://www.statistikdatabasen.scb.se, accessed 27 September 2017). A large increase (72%) was also seen for SAIR incidents from 2012 to 2014, and a smaller for SAIR injuries (19%). While it is possible that the ISA and SAIR numbers reflect an actual increase in injuries/incidents, at least two factors suggest an increased reporting frequency. First, the change in the benefit scheme starting in April 2012, which may have increased propensity to report to ISA as well as likelihood for absence from work (SWEA, 2013). Second, the introduction of the online reporting interface for both ISA and SAIR (anmalaenarbetsskada.se, December 2011) facilitating employers’ reporting. In particular, the increasing numbers for SAIR incidents highlight the importance of simplifying employers’ reporting. That the SAIR injuries did not increase as notably may indicate that severe injuries are decreasing thus offsetting an increased reporting frequency. It is also conceivable that the reporting frequency is not increasing as markedly for this category because employers perceive the ISA reporting as sufficient (and/or the most suitable option of the available reporting forms on the portal page). No increase was seen for PIC major risk cases, although all occupational cases increased by 30% (Schenk et al., 2018). One potential explanation for this overall increase could be the precautionary statements introduced on labelling by the CLP regulation, of which some refer to PICs. However, there are also indications of a general increased awareness of PIC as also non-occupational cases have increased during the investigated period (Schenk et al., 2018). The different purposes of ISA, SAIR and PIC also influence the information collected about the injuries and incidents. Both ISA and SAIR focus on underlying causes, for example pipe-failure, rather than the nature of the chemical hazard, for example sodium hydroxide. This may allow identification of more general preventive measures, but will not allow identification for instance of products with problematic uses. PIC experts, on the other hand, have knowledge and resources at hand to identify the major hazard of the exposure. Hence, the PIC data provide reliable data on the identification of chemicals and products, in addition to route of exposure and a medical severity scoring. The focus of ISA hampers simple identification of injuries involving hazardous substances. Our starting point was cases registered as caused by ‘dusts, gases, fumes or liquids’ (n = 1951). However, after scrutiny of the data for these cases we concluded that at least 37% of these did not involve a chemical injury. Our targeted SAIR excerpt identified 391 injuries and 817 incidents, although future analyses of chemical injuries/incidents using SAIR data could be restricted to only injuries/incidents caused by chemicals as the high proportion of certain kinds of hazardous substances found under other categories (e.g. fire fumes and pharmaceuticals) may reflect concern for other hazards than toxicological. Nevertheless, a comparison of the dates and workplaces recorded in ISA and SAIR injuries shows only 20 cases overlap for 2012 through 2014, which is low also if only taking the SAIR category of chemicals into account (n = 298). As SAIR injuries should be a subset of ISA, this shows that ISA is not specific with regards to identifying cases involving hazardous substances. Possibly employers reporting to ISA perceive certain exposures as more suitable under other categories (e.g. ‘electricity, explosion or fire’ for fire fumes or exploding batteries). Hence, other sorting criteria, based on injury or contact descriptors, as included in the EU level ESAW methodology, may be more relevant for future reviews. Still, some difficulties with identifying chemical accidents will likely remain, as reporting of occupational injuries is highly dependent on coder even when guided by an ESAW compatible reporting form (Molinero-Ruiz et al., 2015; Jacinto et al., 2016). Furthermore, to identify of the nature of chemical hazard in ISA required a non-trivial coding effort by combining information under material agent and injury. For one-fourth of the ISA cases corrosion hazards were identified due to the recorded injuries, for another 41% hazards were not identifiable at all. Similarly, we were unable to connect 44% of ISA cases to any product group. For SAIR, we used the free-text description to identify chemical hazard (possible in 95% of cases), also a time-consuming procedure unfeasible for frequently repeated surveillance efforts. Hence, from the perspective of investigating acute chemical exposures at work, additional information sources with readily available hazard or product information are valuable. The lack of specificity regarding the identity of hazardous substances is connected to the accuracy of exposure data. Both ISA and SAIR are based on employers reporting, and as discussed above how they code accidents/incidents in the notification forms may differ significantly between individuals. Moreover, there may be incentives to not report circumstances accurately or in detail. PIC records also depend on the callers’ account, but trained experts enter the records and ask for the details needed for giving an appropriate advice, a target that also is in the interest of the caller. The lacking specificity and accuracy could partly be addressed by SWEA by adding more guidance on how to report material agents to the online reporting form. Addition of a reporting item with predefined exposure categories for cases involving hazardous substances to the ISA and SAIR reporting schemes could further enhance the specificity and accuracy of these data. The analysis of overlapping dates showed that the three databases largely cover different accidents/incidents. The comparison of hazards and products groups also shows differences between ISA, SAIR and PIC. Cleaning agents was the most common product group among PIC major risk cases and although relatively common also in ISA and SAIR, they were notably less so. This could indicate a systematic under-reporting of incidents with cleaning agents to SWEA. The lower frequency among SAIR incidents implies that cleaning agents are not perceived as having the potential for severe injuries. The hazards identified in our material were generally corrosion or irritation, due to alkali or acid properties. Acids and alkali in cleaning agents have also been frequently implicated for chemical ocular injuries (Tschopp et al., 2015). Suleiman and Svendsen (2014) found that safety data sheets for cleaning agents were inaccurate and/or incomplete, not allowing identification of all hazards or sufficient risk management measures. Furthermore, training of cleaning workers was found ineffective from a chemicals risk management perspective, as providers of training were lacking competence and as chemical health hazards were given comparably low priority (Suleiman and Svendsen 2015). Cleaning agents are used in many sectors, in some of which precarious employments are commonplace, which may decrease propensity to report injuries (Probst et al., 2013; Picchio and van Ours, 2017). Further research on the issue of acute exposures to cleaning agents, the causes and consequences is thus warranted. Turning to the use of PIC data in occupational injury and hazard surveillance, we find that the data offer complementary knowledge but that there are several factors limiting the use of PIC data for surveillance purposes. For instance, while in the present work cleaning agents were identified as an under-recognized hazard, PIC data offer no guidance on which kind of workplaces these hazards occur. Hence, PIC data alone cannot guide SWEA’s workplace inspections, or other kinds of workplace interventions. Furthermore, although a follow-up analysis of ISA and/or SAIR data could provide workplace data, the PIC data may concern a larger range of workplaces than covered by ISA. Nevertheless, by identifying product groups interventions could be targeted to relevant suppliers, who do inform downstream users about proper uses and risk management measures through the safety data sheets. For such identification of injury hazards also the PIC minor and moderate risk cases could be valuable to analyse as they, unlike ISA and SAIR accidents, offer opportunities for experience feed-back from incidents with no to minor injuries (Schenk et al., 2018). Although this is also the case for SAIR incidents, the PIC covers a significantly larger amount of cases; in 2014 the PIC handled 1831 occupational cases, five times as many as we identified for hazardous substances in SAIR incidents (Supplementary Table S2, available at Annals of Work Exposures and Health online). However, for continuous occupational injury surveillance regarding injuries caused by hazardous substances it might be more appropriate to improve the accuracy and specificity of ISA and SAIR reporting. Summary and conclusions None of the three database investigated in the present study can be expected to be complete, that is cover all accidents/incidents taking place at Swedish workplaces. We identified issues regarding specificity for ISA, for which it was challenging to separate out the cases involving hazardous substances. Furthermore, neither ISA not SAIR systematically include data on the identity of the hazardous substance, restricting identification of hazards and chemical products. Accuracy of the recorded data is an issue for all three databases. Both ISA and SAIR depend on employers’ reporting, and hence employers’ interpretation of the notification form as well as their truthfulness and attention to detail. The PIC data are recorded by trained experts who ask callers for the information needed to assess hazards and risks. Due to the nature of the PIC service, callers likely have a larger incentive to be detailed and truthful about involved hazardous substances. The strengths of ISA and SAIR are the recording of demographic and/or workplace data, allowing specific follow-up, and the identification of causes of exposures, such as machine failure, which may help identifying general issues behind accidents with a broad range of hazardous substances. The major strength of the PIC data is the identification of hazard and product group. Our comparisons of injury/incident dates between the different databases indicate significant differences in coverage of the different datasets, hence combining knowledge from all three sources should yield a more comprehensive picture of accidents/incidents at Swedish workplaces. In the present study, the comparison of chemical hazards and product groups represented in the different datasets identified cleaning agents as an under-recognized workplace hazard by SWEA data. Given the lack of specificity of SWEA data, periodical reviews of occupational cases handled by the Swedish PIC provide complementary knowledge regarding acute exposures to hazardous substances and may identify needs for further research and/or additional risk management efforts. Occupational injury reporting differs between countries and regions (Hämäläinen et al., 2006). Hence, transfer of our findings to other countries need to take national injury reporting systems and PIC coverage into account. Nevertheless, we find that co-operation between PICs and work environment authorities would improve the knowledge basis on acute chemical exposures at work, which in turn could be used to improve of the working environment. Supplementary Data Supplementary data are available at Annals of Work Exposures and Health online. Funding Funding for this research was provided by AFA Insurance (dnr 130288), an organization owned by Sweden’s labour market parties. The research at Swetox was supported by the Stockholm County Council, the Knut & Alice Wallenberg Foundation, and the Swedish Research Council FORMAS. Acknowledgements The authors thank Karin Feychting and Anita Annas, The Swedish Poisons Information centre, and Kjell Blom, the Swedish Work Environment Authority, for assistance with data retrieval and informative discussions about the respective data sources. Two anonymous reviewers are gratefully acknowledged for insightful comments and suggestions. The authors declare no conflict of interest relating to the material presented in this article. Its contents, including any opinions and/or conclusions expressed, are solely those of the authors. References Blanc PD , Olson KR . ( 1986 ) Occupationally related illness reported to a regional poison control center . Am J Public Health ; 76 : 1303 – 7 . Google Scholar CrossRef Search ADS PubMed Eurostat . ( 2013 ) European Statistics on Accidents at Work (ESAW) summary methodology . Available at http://ec.europa.eu/eurostat/documents/3859598/5926181/KS-RA-12-102-EN.PDF/56cd35ba-1e8a-4af3-9f9a-b3c47611ff1c. Accessed 27 September 2017 . Fagan KM , Hodgson MJ . ( 2017 ) Under-recording of work-related injuries and illnesses: an OSHA priority . J Safety Res ; 60 : 79 – 83 . Google Scholar CrossRef Search ADS PubMed Gravseth HM , Wegeland E , Lund J . ( 2003 ) Underrapportering av arbeidsskader til Arbeidstilsynet [Underreporting of occupational injuries to the labour inspectorate] . Tidskr Nor Lægefor ; 15 : 2057 – 59 . In Norwegian, English abstract available. Hämäläinen P , Takala J , Saarela KL . ( 2006 ) Global estimates of occupational accidents . Saf Sci ; 44 : 137 – 56 . Google Scholar CrossRef Search ADS Hinnen U , Hotz P , Gossweiler B et al. ( 1994 ) Surveillance of occupational illness through a national poison control center: an approach to reach small-scale enterprises ? Int Arch Occup Environ Health ; 66 : 117 – 23 . Google Scholar CrossRef Search ADS PubMed Jacinto C , Aspinwall E . ( 2004 ) A survey on occupational accidents’ reporting and registration systems in the European Union . Saf Sci ; 42 : 933 – 60 . Google Scholar CrossRef Search ADS Jacinto C , Santos FP , Guedes Soares C et al. ( 2016 ) Assessing the coding reliability of work accidents statistical data: how coders make a difference . J Safety Res ; 59 : 9 – 21 . Google Scholar CrossRef Search ADS PubMed Leigh JP , Marcin JP , Miller TR . ( 2004 ) An estimate of the U.S. Government’s undercount of nonfatal occupational injuries . J Occup Environ Med ; 46 : 10 – 8 . Google Scholar CrossRef Search ADS PubMed Litovitz T , Oderda G , White JD et al. ( 1993 ) Occupational and environmental exposures reported to poison centers . Am J Public Health ; 83 : 739 – 43 . Google Scholar CrossRef Search ADS PubMed Meulenbelt J , de Vries I . ( 1997 ) Acute work-related poisoning by pesticides in The Netherlands; a one year follow-up study . Przegl Lek ; 54 : 665 – 70 . Google Scholar PubMed Molinero-Ruiz E , Pitarque S , Fondevila-McDonald Y et al. ( 2015 ) How reliable and valid is the coding of the variables of the European Statistics on Accidents at Work (ESAW)? A need to improve preventive public policies . Saf Sci ; 79 : 72 – 9 . Google Scholar CrossRef Search ADS Olson DK , Sax L , Gunderson P et al. ( 1991 ) Pesticide poisoning surveillance through regional poison control centers . Am J Public Health ; 81 : 750 – 3 . Google Scholar CrossRef Search ADS PubMed Persson HE , Sjöberg GK , Haines JA et al. ( 1998 ) Poisoning severity score. Grading of acute poisoning . J Toxicol Clin Toxicol ; 36 : 205 – 13 . Google Scholar CrossRef Search ADS PubMed Picchio M , van Ours JC . ( 2017 ) Temporary jobs and the severity of workplace accidents . J Safety Res ; 61 : 41 – 51 . Google Scholar CrossRef Search ADS PubMed Probst TM , Barbaranelli C , Petitta L . ( 2013 ) The relationship between job insecurity and accident under-reporting: a test in two countries . Work Stress ; 27 : 383 – 402 . Google Scholar CrossRef Search ADS Rappin CL , Wuellner SE , Bonauto DK . ( 2016 ) Employer reasons for failing to report eligible workers’ compensation claims in the BLS survey of occupational injuries and illnesses . Am J Ind Med ; 59 : 343 – 56 . Google Scholar CrossRef Search ADS PubMed Rosenman KD , Kalush A , Reilly MJ et al. ( 2006 ) How much work-related injury and illness is missed by the current national surveillance system ? J Occup Environ Med ; 48 : 357 – 65 . Google Scholar CrossRef Search ADS PubMed Rubenstein H , Bresnitz EA . ( 2001 ) The utility of Poison Control Center data for assessing toxic occupational exposures among young workers . J Occup Environ Med ; 43 : 463 – 6 . Google Scholar CrossRef Search ADS PubMed SCB . ( 2003 ) Informationssystemet om arbetsskador och undersökningen om arbetsorsakade besvär. En jämförande studie. [Information system on occupational injuries and the work-related disorders survey. A comparative study] Report 2003 :5. p. 44 . Available at http://www.scb.se/statistik/AM/AM0602/1995I00/AM0602_1995I00_AM76ST0305.pdf. Accessed 27 September 2017 . Schenk L , Feychting K , Annas K et al. ( 2018 ) Records from the Swedish Poisons Information Centre as a means for surveillance of occupational accidents and incidents with chemicals . Saf Sci; 104: 269–75 . Sudakin DL , Power LE . ( 2007 ) Pyrethrin and pyrethroid exposures in the United States: a longitudinal analysis of incidents reported to poison centers . J Med Toxicol ; 3 : 94 – 9 . Google Scholar CrossRef Search ADS PubMed Suleiman AM , Svendsen KV . ( 2014 ) Are safety data sheets for cleaning products used in Norway a factor contributing to the risk of workers exposure to chemicals ? Int J Occup Med Environ Health ; 27 : 840 – 53 . Google Scholar CrossRef Search ADS PubMed Suleiman AM , Svendsen KV . ( 2015 ) Effectuality of cleaning workers’ training and cleaning enterprises’ chemical health hazard risk profiling . Saf Health Work ; 6 : 345 – 52 . Google Scholar CrossRef Search ADS PubMed Swedish Work Environment Authority (SWEA) . ( 2013 ) Arbetsskador 2012. Workrelated injuries 2012 . p. 102 . In Swedish. Available at https://www.av.se/globalassets/filer/statistik/arbetsmiljostatistik-arbetsskador-2012-rapport-2013-01.pdf. Accessed 27 September 2017 . Swedish Work Environment Authority (SWEA) . ( 2014 ) Work-related disorders 2014. Report 2014 :4. Available at https://www.av.se/globalassets/filer/statistik/arbetsmiljostatistik-arbetsorsakade-besvar-rapport-2014.pdf. Accessed 27 September 2017 . Swedish Work Environment Authority (SWEA) . ( 2015 ) Årsredovisning 2014. Yearly report 2014 . p. 52 . In Swedish. Available at https://www.av.se/globalassets/filer/om-oss/arsredovisningar/arsredovisning-arbetsmiljoverket-2014.pdf. Accessed 27 September 2017 . Trueblood AB , Shipp EM . ( 2017 ) Characteristics of occupational pesticide exposures reported to poinsin control centers in Texas, 2000–2015 . Arch Environ Occup Health ; 12 : 1 – 8 . Google Scholar CrossRef Search ADS Tschopp M , Krähenbühl P , Tappeiner C et al. ( 2015 ) Incidence and causative agents of chemical eye injuries in Switzerland . Clin Toxicol (Phila) ; 53 : 957 – 61 . Google Scholar CrossRef Search ADS PubMed United States Government Accountability Office (US GAO) . ( 2012 ) Workplace safety and health: better OSHA guidance needed on safety incentive programs . GAO-12– 329. Available at http://www.gao.gov/products/GAO- 12–329. Accessed 27 September 2017 . Woolf A , Alpert HR , Garg A et al. ( 2001 ) Adolescent occupational toxic exposures: a national study . Arch Pediatr Adolesc Med ; 155 : 704 – 10 . Google Scholar CrossRef Search ADS PubMed Ziqubu-Page T , Forrester MB . ( 2016 ) Adolescent workplace exposures reported to Texan poison centers . Int J Adolesc Med Health . [Epub ahead of print] .doi: 10.1515/ijamh-2016-0057 © The Author(s) 2018. Published by Oxford University Press on behalf of the British Occupational Hygiene Society. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Annals of Work Exposures and Health (formerly Annals Of Occupational Hygiene)Oxford University Press

Published: Feb 28, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off