Table 2.
Multivariate techniques used to analyse IMIS exposure data.
| Publication | Agent and setting | Exposure metric | Variables studied | Analytical approach |
|---|---|---|---|---|
| Froines et al. 1991 | silica | median severity by inspection | industry, number of employees, union status, inspection type | logistic regression, response is inspection specific median severity being greater than 1 |
| Gomez 1997 | three subsets: lead in battery manufacturing, perchloroethylene in dry cleaning, iron oxide in welders | concentration, company specific mean concentration, probability of being greater than a specified value | job description, number of employees, union status, year, scope of inspection, type of inspection | for each dataset: linear multiple regression of company specific log-transformed mean concentrations, linear multiple regression of individual log-transformed concentrations with within inspection correlation, logistic regression of the probability for a measurement being greater than the dataset specific 75th percentile of exposure levels |
| Linch et al. 1998 | silica | proportion of workers associated with a fixed severity | year, number of employees | linear model with response the transformed site specific proportion of workers exposed as a function of industry, year and number of employees |
| Teschke et al. 1999 | wood dust | concentration | year, job description, number of employees, inspection type | linear multiple regression of individual log-transformed concentrations with within-company correlation |
| Melville and Lippman 2001 | three subsets: Asbestos abatement, toluene in auto-repair bodyshops, formaldehyde in embalmers | concentration, company specific mean concentration, probability of being greater than a specified value | job description, number of employees, union status, year, scope of inspection, type of inspection | for each dataset: linear multiple regression of company specific log-transformed mean concentrations, linear multiple regression of company specific log-transformed mean concentrations weighted by associated variances, linear multiple regression of individual log-transformed concentrations with within-inspection correlation |
| Coble et al. 2001 | several agents in the pulp and paper industry | concentration | industry, job description, year | linear regression of log-transformed concentrations on year of measurement |
| Lurie and Wolfe 2002 | hexavalent chromium | concentration, number of measurements, citations | year, industry, inspection type, inspection conducted by federal or state agency | univariate linear regressions and rank sum tests |
| Hennerberger et al. 2004 | beryllium | companies with most recent inspection associated with beryllium levels greater than 0.1 or 0.5 μg/m3 | year, number of employees | linear multiple regression with response the transformed site specific proportion of workers exposed as a function of SIC, year and number of employees |
| Middendorf 2004 | noise | several noise exposure metrics | year, number of employees, | linear regression for noise exposure versus year and general linear model for noise level versus number of employees + year |
| Okun et al. 2004 | lead | probability of a measurement exceeding the PEL | year, region, number of employees, union status, inspection type | SIC specific logistic regression with correlation within inspection (fit using generalized estimating equations), response is individual sample result being greater than PEL |
| Yassin et al. 2005 | silica | concentration | year, industry, job description, inspection type | non parametric regression to test the hypothesis of similar mean exposure in all industries, autoregressive ARMA (2) model with errors correlated with previous and following time periods. Covariates included year, industry, and inspection type |
| Flanagan et al. 2006 | silica in the construction industry | concentration | several exposure determinants not documented in IMIS+ year | linear multiple regression of individual log-transformed concentrations |
| Lavoué et al. 2008 | formaldehyde | concentration | inspection type, sample type (short-term, TWA), season, industry, year, state, outside temperature, | linear multiple regression of individual log-transformed concentrations with within-inspection correlation, TOBIT models |
| Lavoué et al. 2011 | formaldehyde | concentration | data source, year, sample type (short-term versus TWA), industry | linear multiple regression of individual log-transformed concentrations with within-inspection correlation, TOBIT models, multimodel inference as the model selection framework |
| Hamm and Burstyn 2011 | beryllium | evaluation leading to beryllium level greater that 0.1 μg/m3 or 0.5 μg/m3 | industry, job description, measurement being TWA, year | Poisson multiple regression with random sample effect |
| Henn et al. 2011 | lead | percent of samples over the PEL | industry, time period, region, number of employees, federal/state plan, union status, inspection type, advance notice of inspection, presence of employee representative, employees interviewed during inspection | logistic regression, response is the probability of a measurement being greater than the PEL |