|Home | About | Journals | Submit | Contact Us | Français|
V. Bonneterre1, D. J. Bicout1, L. Larabi2, A. Maitre1, P. Tubert‐Bitter3, R. de Gaudemaris1. 1Laboratoire Environnement et Prédiction de la Santé des Populations – TIMC, Grenoble, France; 2Occupational Medicine Department, Grenoble University Hospital; 3Inserm U 472, Hôpital Paul Brousse, Villejuif
ObjectivesEarly detection of new disease–occupational nuisance associations is an important health issue. The French national occupational disease surveillance and prevention network database (RNV3P, for Réseau National de Vigilance et Prévention des Pathologies Professionnelles) was created in 2001 to monitor occupational and environmental diseases. One of its aims is to develop statistical methods or models to detect potentially emerging associations between diseases and nuisances. The aim of this work is to test data mining methods used in pharmacovigilance in order to identify potential emerging disease–nuisance associations (DNA) in the RNV3P.
MethodsApplication to the RNV3P database of safety data mining methods used in pharmacovigilance. These methods analyse drug–event pairs which are the equivalent of the disease–occupational nuisance pairs or associations included in the RNV3P and generate a statistical signal when there is a discrepancy between the true number of pairs reported and the expected number for this same pair in the database. DNA which do not yet give entitlement to compensation and which generate a signal with these methods are pointed out as “potentially emergent” associations.
ResultsThe 24785 reports of the RNV3P were grouped into 1344 different DNA reported more than twice, of which 422 did not give entitlement to compensation by the social security system. Among these DNA, 162 were “potentially emergent” and generated a signal, in particular eight associations involving cancer. These formerly “unrecognised” DNA, reported several times by the network, originate either from potential toxicological problems (associations where the relationship of cause to effect is not yet established), or from problems of social recognition: work‐related health disorders which do not yet give entitlement to compensation.
ConclusionThis work is a preliminary step towards automated systematic analysis of RNV3P data in order to detect potentially emerging DNA. Its aim is a dual one: to reveal, as early as possible, the new occupational risks which can be detected from the network database (hypothesis‐generating function) and to emphasise the role of the network in forming expert opinion, as the hypotheses generated are transmitted to the experts of the occupational disease centres who assess their pertinence.
Key wordsoccupational network or database; signal detection; data mining
R. McNamee1, Y. Chen2, L. Hussey1, R. Agius1. 1University of Manchester; 2Alberta Cancer Board
ObjectivesIn an existing surveillance scheme for occupational disease, occupational physicians reported new cases diagnosed in one, randomly chosen month of every year. Annual incidence was then inferred as reported incidence multiplied by 12. This calculation might be invalid if reporting frequency influences reporter behaviour. We wanted to test the hypothesis that estimated annual incidence rates differ systematically when reporting is every month (12/12) compared to 1 month a year (1/12).
MethodsA subgroup of existing reporters were invited to take part in a randomised controlled cross‐over trial where there would be random allocation to report either 12/12 or 1/12 in 2004; reporters would then cross over to the other reporting pattern in 2005. Data for each reporter consisted of up to 13 monthly counts of new cases of occupational disease. The ratio of each count to the reporter's monthly average prior to the study – the “standardised case count” – was calculated and compared between trial arms and years. The counts were analysed using a two‐level Poisson regression model – with offset term equal to the reporter's pre‐study average count – and reporting frequency and year as predictor variables.
ResultsNinety‐seven people were asked to take part; 63 (65%) accepted and were randomly allocated to either: report 12/12 in 2004 and 1/12 in 2005 (n=32); or the reverse (n=31). Overall, 87% (712/819) of the possible monthly reports were returned. The “standardised case count” was higher when reporting 1/12 compared to 12/12 in both years (means 1.23 vs 0.82, respectively, in 2004 and 1.07 and 0.77, respectively, in 2005). From the Poisson regression analysis, annual incidence based on 1/12 reporting was estimated to be 26% (95% CI 11% to 42%, p<0.001) higher than for 12/12 reporting. Incidence was estimated to have fallen by 13% (95% CI 2% to 23%, p<0.025) from 2004 to 2005.
ConclusionThis randomised controlled trial provides unambiguous evidence that reporting frequency influences estimated incidence. Possible explanations are that 1/12 reporting overestimates, or 12/12 reporting underestimates incidence, or both. Future analyses, to be presented, will address possible fatigue in 12/12 reporting, and will help decide which type of reporting gives the most accurate estimate of incidence.
Key wordssurveillance methods; incidence estimation; reporting frequency
M. Concha1, J. Labbe2, J. Giaconi1. 1Asociacion Chilena de Seguridad; 2UNAB
ObjectivesInformation on occupational health needs to be pooled to develop more systematic evidence of the importance of the field in Latin American countries. Indicators that are easily understood for the population and stakeholders should be a tool to increase resource allocation in the field. The purpose of this study is to estimate the life expectancy with disability due to injuries and diseases due to work to demonstrate the importance of workers' health.
MethodsSullivan's method was used to calculate the years lived with disability. The method uses a standard life table for the population and the observed prevalence of disability by age to estimate years lived with disability. The life table was developed using government data (Chile, 2004). Disability prevalence was gathered from survey data for disability caused by specific conditions. To estimate the indicator, Sullivan's method uses the life table functions.
ResultsIn Chile, the life expectancy at birth in 2004 was 77.5 years for the total population. The gap among sexes was almost 6 years. The total years lived with disability at birth (for any specific cause) were 14.2 years for males and 19.9 years for females. The years lived with disability due to occupational conditions and injuries range from 1.3 years for males and 0.6 years for females. Health problems due to occupation correspond to the third leading cause, causing almost 10% of the total years with disability. For females only 3% of the years living with disability are related to occupational injuries or diseases.
ConclusionTo our knowledge, this is the first study in Latin America to explore the impact of disability due to occupation on population health expectancy. Empirical evidence about frequency and severity of condition related to occupation suggests that our results may underestimate the impact of occupation on health expectancy. However, our results highlight the importance of using new indicators in the occupational field to demonstrate the importance of occupational conditions for the well‐being of the population. Further research and refinement of the methodology is required.
Key wordsindicator; disability; health expectancy
M. Valenty1, J. Homere1, M. Pubert1, A. Touranchet2, C. Ha1, E. Imbernon1. 1Institut de Veille Sanitaire; 2Ministère du Travail
ObjectivesIn France, financial compensation for an occupational disease relies on tables that are based on presumption of causality and define the necessary and sufficient conditions. Work related diseases (WRD) are theoretically all the diseases with an occupational origin but not compensated. The declaration of putative WRD is an obligation for all physicians and is aimed at updating the tables of occupational diseases. Actually, however, physicians seldom notify the diseases they think may be linked with work conditions or occupational exposures. Moreover, these notifications cannot be used for epidemiologic surveillance due to lack of data on the source population. Therefore, the InVS has developed a protocol for WRD surveillance. The main purpose is mainly to estimate the prevalence of WRD in the working population according to age, gender and economic sector. In 2007, SUMATRAS was implemented in six regions.
MethodsThe registration system was implemented in the Pays de la Loire region in 2003, and was extended to three others regions in 2006. Actually, it is based on a 2‐week registration period repeated every 6 months through a network of volunteer occupational physicians. These physicians notify the WRD they have observed during the annual visit of workers. In order to estimate the prevalence rates, we collected data on age, gender and economic sector of all the workers visited by the same physicians during the same period.
ResultsIn 2006, 800 occupational physicians (about 35%) volunteered to take part in the four regions. The economic sectors of the regions were quite well represented. The WRD prevalence varied from 5% to 8% according to the region and the period. Musculoskeletal disorders represented the main cause of diseases; the second cause was mental disorders.
ConclusionBetter registration was observed during these 2‐week registration periods in comparison with the other regions where this program was not implemented. Regularly repeated, these short periods of registration will provide useful data on the frequency of pathological manifestations associated with work, as well as an estimate of the extend of undeclared compensatible WRD.
Key wordswork‐related diseases; surveillance; prevalence