|Home | About | Journals | Submit | Contact Us | Français|
We examined the effectiveness of preventive interventions against occupational injuries (preferential action plans [PAPs]) developed by Spanish regional governments starting in 2000.
We included 3,252,028 occupational injuries with sick leave due to mechanical causes occurring between 1994 and 2004 in manufacturing and private service companies. Time trends for occupational injury rates were estimated before and after implementation of PAPs in each region, with a control group defined for those regions in which no PAPs were implemented (e.g., Galicia, Madrid, and Cataluña). We determined annual change percentages and their 95% confidence intervals (CIs) through a negative binomial regression model. Regions were grouped into three categories according to formal quality of their PAPs.
The regions with the best PAPs (Andalucia, Aragon, Valencia, and Murcia) showed annually increasing occupational injury rates (2.3%, 95% CI –2.5, 7.4) before implementation of PAPs. After PAPs were implemented, occupational injury rates decreased significantly to –7.4% (95% CI –10.2, –4.5). Similar results were also found for regions with PAPs of lower quality and even for regions that didn't implement a PAP (control group). These results did not vary substantially in stratified analysis by gender, age, type of contract, or length of sick leave.
PAPs are not related to a general decline in occupational injury rates in Spain starting in 2000. Reinforcement of Spanish health and safety regulations and labor inspection activities since 2000, resulting from a social agreement between central government and social agents, remains an alternative hypothesis requiring additional research.
In Spain, as well as in other countries, occupational injury is an important public health problem, and it is the main concern for occupational health and safety public policies.1–3 However, the body of evidence for evaluating the effectiveness of occupational health and safety interventions is limited, generally due to methodological weakness.4
After a serious accident occurred in a shipyard of Valencia, Spain, in 1997, leaving 18 workers dead,5 some Spanish regions were prompted to develop specific preventive interventions against occupational injuries (called preferential action plans [PAPs]), which were focused on those companies with a high number of occupational injuries. It was observed that approximately 40% of the occupational injuries occurred at 2% of the companies.6 An agreement between central government and social agents (e.g., trade unions and employers) was reached in 1998 to reinforce the labor inspectorate actions and public prosecutor by a special attorney from the central Spanish administration.
PAPs work primarily by (1) visiting those companies with a high number of occupational injuries; (2) finding out if those companies fulfill legal requirements concerning preventive measures, mainly related to safety (e.g., machinery, equipment, tools, devices, and clean spaces); (3) offering solutions; and (4) establishing deadlines to solve detected faults. Between 1999 and 2004, the labor administration of the majority (14 out of 17) of the Spanish regions developed PAPs in their own territories, with some variations among them.7 PAPs were especially focused on nonfatal traumatic occupational injuries, because about one million of these injuries occur annually in the whole country and because fatal traumatic occupational injury rates in Spain have been declining since 1992,8 following a similar trend in other developed countries.9–12
This study examined for the first time the effectiveness of PAPs on occupational injury trends in Spain. This may be an example for other countries that want to assess their public policies regarding occupational injury prevention.
This study analyzed occupational injury rates by Spanish regions between 1994 (the year before the European Framework Directive 89/391 was translated into Spanish13) and 2004, comparing the periods before and after each region implemented its own PAP (1999 in Aragon; 2000 in Valencia, Murcia, Pais Vasco, Rioja, Asturias, Navarra, Cantabria, Extremadura, Castilla-La Mancha, and Canarias; 2001 in Andalucia; 2002 in Baleares; and 2003 in Castilla-Leon).
We reviewed available documentation about PAPs, supplied by each region's labor administration, according to 12 explicit and defined criteria, previously agreed to and scored by the researchers.7 The PAP must:
These criteria were applied independently by two researchers; a third researcher made the decision in case of disagreement between the first two. The total quality score for each PAP was obtained by adding the points corresponding to each previous criterion (maximum total quality score = 34 for a PAP accomplishing all previous criteria; minimum total quality score = 0 for a PAP accomplishing any of these previous criteria).
An analysis by Spanish regions has been published previously.14 In this article, regions were grouped into three categories so that we could assess a possible gradient between the quality of the PAP and its impact on occupational injury rates. Regions with a PAP total quality score of >25 points were considered “excellent” (e.g., Valencia, Andalucia, Aragon, and Murcia); regions with a PAP total quality score between 25 and 15 points were considered “good” (e.g., Pais Vasco, Rioja, and Castilla-Leon); and regions with a PAP total quality score of <15 points were considered “fair” (e.g., Baleares, Asturias, Navarra, Canarias, Cantabria, Extremadura, and Castilla-La Mancha). Three remaining Spanish regions (Madrid, Galicia, and Cataluña) were used as a control group because they did not develop any recognized PAP during the study period. In this control group, occupational injury trends were analyzed before and after 2000, because this was the year when PAPs were first implemented in the majority of Spanish regions.
For this analysis, nonfatal occupational injuries occurring on the worksite, causing at least one day of sick leave and attributable to mechanical causes, were selected. These included injuries due to falling from a height or at the same level, being hit by falling objects, falling objects being manipulated, being struck by projectile fragments or particles, being trapped by or between objects, being trapped by vehicles or machinery rollover, stepping on or tripping over an object, being struck against a stationary object or by a moving object, and being struck by an object or tool. All fatal occupational injuries, nonfatal occupational injuries attributable to nontraumatic causes (e.g., heart attack or stroke), commuting accidents, and other nonfatal injuries such as overexertion were excluded. Furthermore, only occupational injuries occurring in manufacturing and private service companies, representing 23% and 47% of the Spanish workforce in 2004, respectively,15 were considered for PAP actions.PAPs were generally not considered in public service companies, as they have special occupational health and safety regulations, nor in construction or primary sector industries (e.g., agriculture, fishing, and mining) due to difficulties related to access and intervention in these mostly small and disperse companies.
This analysis included 3,252,028 occupational injuries that met inclusion criteria as registered by the Ministry of Work and Social Affairs between 1994 and 2004. The Labor Force Surveys from the Spanish Institute of Statistics, providing data on salaried workers in Spain for the study period, were used as the best available denominator for the analysis. The Spanish Labor Force Survey is conducted quarterly on a representative sample of about 250,000 people, including salaried, autonomous, and cooperative workers. To estimate occupational injury rates in this study, we used salaried workers as the best approach for total population at risk.15 A previous study showed that the Labor Force Survey was a valid source in providing denominators for estimating occupational injury rates by age, gender, region, and economic activities in Spain.16
Annual occupational injury rates were calculated for each category of regions (with excellent, good, and fair PAP and control groups), and annual change percentages and their 95% confidence intervals (CIs) were estimated assuming a negative binomial distribution17 on yt, the number of nonfatal occupational injuries in year t, with the following log-lineal mean:
In this regression model, Nt indicates person-years in t and τ is the year during which time trend changes. The expression (t–τ)+ is equal to t–τ when t>τ and equal to 0 when t≤τ. The annual percentage changes were estimated as 100 × [exp(β1)–1] for the period before the PAP was implemented and as 100 × [exp(β1+β2)–1] for the period after PAP implementation. Negative values showed decreasing trends in the incidence of occupational injuries, and positive values showed increasing trends. The analysis was conducted using the R program.18
Stratified analysis was conducted by gender, age group (<25 years, 25–34 years, 35–54 years, and ≥55 years), type of contract (permanent or temporary), and length of sick leave (≤15 days or >15 days). All were variables that could potentially interact19 as effect modifiers, with the main relationship between PAP and occupational injury rate trends. Workers with temporary contracts have shown a higher risk than workers with permanent contracts of having occupational accidents.20 Length of sick leave was also included in the analyses because of the suspicion that some non-occupational injuries could be declared as occupational injuries, especially for those with <16 days of sick leave, as companies have to pay something for these first 15 days in case of a nonoccupational injury and nothing in case of an occupational injury.
In 2000, occupational injury rate trends began to decline uniformly in the three categories of Spanish regions (with excellent, good, and fair PAP and control group scores) in manufacturing industries as well as in private service companies (Figure 1). In the category of regions with excellent PAP scores, nonfatal traumatic mechanical occupational injury rates in manufacturing and private service companies increased by a mean of 2.3% (95% CI –2.5, 7.4) annually before their PAPs were implemented (Figure 2). A similar result was found in the control group and in groups with good and fair PAP scores. Conversely, after the PAPs were implemented, injury rates in the group with excellent PAP scores decreased significantly (–7.4%, 95% CI –10.2, –4.5). Similar results were also found for the groups with good PAP scores (–9.0%) and fair PAP scores (–9.5%), as well as for the control group (–10.1%), even showing slightly higher declines with decreasing quality of PAP and lack of it, although the 95% CIs mostly overlapped. Very similar patterns were found when annual percentage changes were estimated individually for manufacturing companies and for private service companies (data not shown).
The Table shows an increasing trend in occupational injuries before implementing PAPs and a decreasing trend after implementing PAPs in all three categories of regions with PAPs, and in the control group. This pattern was also found when male and female workers were analyzed separately. Also, the pattern was very similar in the analysis by age, although for workers younger than 25 years of age, the increase of occupational injury rates before implementing a PAP was markedly high in all groups: 8.8% (excellent), 6.3% (good), 8.8% (fair), and 6.2% (control group).
According to type of contract, the pattern was slightly different between permanent and temporary workers: rates rose significantly before implementing a PAP in all groups only among temporary workers (9.4% [excellent], 6.2% [good], 9.4% [fair], and 7.8% [control group]) and decreased significantly after implementing a PAP in both temporary and permanent workers. Finally, occupational injuries causing sick leaves lasting up to 15 days rose significantly before implementing a PAP (3.9% [excellent], 3.7% [good], 5.1% [fair], and 3.9% [control group]), but not those injuries causing >15 days of sick leave. However, after PAP implementation, occupational injury rates decreased significantly in both categories of sick leave duration (Table).
Our analysis showed no relationship between plans developing specific interventions for the prevention of occupational injuries (e.g., PAPs) and trends of mechanical and nonfatal occupational injuries in Spain. These results did not vary substantially when analyses were performed accounting for gender, age, type of contract, and length of sick leave of injured workers.
As such, the question about what explains the generalized change of occupational injury rate trends in 2000 in Spain remains. Mechanical occupational injury rates in manufacturing and private service companies rose from 49.6 per 1,000 workers in 1994 to 54.4 per 1,000 workers in 2000 (an increase of 9.7%) and abruptly dropped between 2000 and 2004 (30.7 per 1,000, a decrease of 43.7%). Similar trends have been observed in all Spanish regions.14,21 An alternative explanation for these changes is that agreements reached between central government and social agents mobilized health and safety inspection and control, enforcing regulations and penalties in the whole country. Labor inspectors work all across the country and have the authority to make unannounced inspections, investigate compliance with occupational safety and health regulations, and levy fines and penalties in the event of noncompliance. Conversely, PAPs fall under the responsibility of Spanish regional governments, and their staff members are not allowed to visit companies at any time or to investigate noncompliance with regulations, as this is an exclusive function of Spanish labor inspectors.
Another potential explanation for our results could be that employers and Spanish insurance companies (Mutuas) have introduced some changes in reporting occupational injury cases since the labor inspection increased its control activities. There is evidence of the impact of notification regulation changes on occupational injury trends in other countries,22 and it could be the case in Spain as well. Some Spanish companies could have been tempted to reduce occupational injury notification as a dishonest way to get out of the “riskier companies” lists included in the PAP. However, this hypothesis is unlikely because the trend has changed in all autonomous communities, and at the same time and in a consistent way between 2000 and 2004. Additionally, time trend rate falling continued in 2005 and 2006.23
Finally, a change in workforce composition from high-risk to low-risk industries could explain a reduction in injury rates, as it has been shown in relation with fatal occupational injuries in other countries.9 However, between 1994 and 2004, employment increased in all Spanish regions by 30% to 40% in manufacturing companies, and 70% to 90% in private service companies.15 Moreover, such a change is expected to modify injury trends slowly over a relatively large time period, but not as sharply as seen in 2000. On the other hand, for our analysis only occupational injuries occurring in some specific industries (e.g., manufacturing and private service companies) were selected, and working conditions and safety measures in this category of occupations probably have not experienced strong changes during the relatively short study period.
Also, observed occupational injury rate trends seem not to be a consequence of economic fluctuations. Some authors postulate a relationship between the periods of economic growth in a country (i.e., rises in workforce and expanded work hours) with increases in occupational injuries.24 However, Spain has shown an annual rate of economic growth of about 3% since 2000,25 and at the same time occupational injury rates were clearly declining.
This study had some limitations. We didn't have exact estimations of the quality of our data (registered occupational injuries and working populations at risk), but the quality of these records likely did not significantly change during the study period. Also, time trend analysis was only based on 11 years of observation, which is a relatively short period of time. However, the use of binomial regression models took into account potential overdispersion, and 95% CIs gave a measure of variability in the estimations of annual percentage changes. On the other hand, we have assessed the impact of each PAP on the total number of occupational injuries occurring in each region, while the PAPs were focused on specific companies. Unfortunately, identification of companies affected by PAPs was not possible. Further research is needed to assess the specific impact of PAPs on affected companies. However, our approach has additional interest as it takes into account the possible general effect of PAPs, even on companies that were not included for PAP actions.
A major strength of the study was its quasi-experimental design, which allowed us to evaluate the effectiveness of public policies with regions grouped according to different levels of exposure (excellent, good, and fair PAPs) and a control group (without PAPs). This particular design highly supports the validity of observed results and makes them particularly interesting for decision makers.26 Another strong point of the study was the high number of observations and their specificity: only occupational injuries more directly related to interventions developed as a consequence of PAPs were included in the analysis.
This study provides useful evidence for evaluating the effectiveness of occupational health and safety policies. Occupational injury prevention activities developed in Spain, mostly focused on visits and advice for riskier companies, do not seem to be related to the observed generalized decline in occupational injuries in Spain since 2000. Social agreement between the central government and social agents (e.g., trade unions and employers), which was reached in 1998 with subsequent reinforcement of labor inspection actions, could be an alternative explanation for this decline, but additional research would be needed to confirm or reject this hypothesis.
This work was conducted in the Spanish Occupational Health Observatory and was partially funded by a grant from the Ministry of Work and Social Affairs (FIPROS 2005-39).