190 likes | 348 Views
A job exposure matrix for physical load: Dutch experiences. Lex Burdorf Department of Public Health University Medical Center Rotterdam. When to consider a job exposure matrix ?. JEM as only possibility: - no information on physical load available at individual level
E N D
A job exposure matrix for physical load:Dutch experiences Lex Burdorf Department of Public Health University Medical Center Rotterdam
When to consider a job exposure matrix ? JEM as only possibility: - no information on physical load available at individual level - jobs have distinct exposure profiles with combined exposure to several determinants of physical load JEM to be considered: - assessment at individual level with large within-worker variance (classical vs Berkson's error) - historical trends
Description of jobs Hierarchy in questions on occupational job title: - what is your job ? (generic description) - in which company / industry do your work ? (generic description) - what are your daily activities ? (generic description) Hierarchy in data collection on occupational job titles: General questionnaire: 1. What is your job ? 2. Which branch of industry ? Job-specific questionnaire: 3. Frequency and duration of specific activities within jobs ? Tielemans et al Occup Environ Med 1999;56:145-51.
What is your job ? Problems: * Too generic description, eg top 10 jobs in community-based study among pregnant women in Rotterdam 2004-2006 * description not specific enough, e.g. nurse, agricultural worker
Job description among women Top 10: administrative employee 389 8.0% teacher 172 3.5% nursery, kindergarten worker 153 3.1% shop assistent 141 2.9% secretary 139 2.9% lawyer, barrister 139 2.9% physician 123 2.5% cleaner 123 2.5% nurse 108 2.2% cashier 87 1.8% 1574 32.3%
What is your job ? Problems: * Too generic description, eg top 10 jobs in community-based study among pregnant women in Rotterdam 2004-2006 * description not specific enough, e.g. nurse, agricultural worker * 'interesting' descriptions, eg sandwich-designer, visual presenter Experiences: * problems larger among jobs in industry (eg operator) * large variability within job titles in self-reported exposure to determinants of physical load
What is your job ? Problems: * large variability within job titles in self-reported exposure to determinants of physical load (4 point scales) Nursery / kindergarten worker: often / always Standing 38% Sitting 85% MMH > 5 kg 50% MMH > 25 kg 6% - variation within job ? - question not valid ? - interpretation: average exposure within job ?
Which company / industry do your work ? Problems: * company name is not always very informative (e.g large companies) * description not specific enough, e.g. government Experiences: * nurse in hospital is not informative without knowing the specific department and specific activities
What are your daily activities ? Problems: * no information on frequency and duration * no standardized description possible, hence, sometimes difficult to interpret Experiences: * cannot be linked to exposure pattern, useful additional information for job classification
Classification of occupational job titles Coding systems: - current classification; education, job, activities, competencies 5 digits, first digit 5 classes, 3 digits job group, 5 digits job title - old classification; jobs within specific industries 4 digits, 2 digits branch of industry, 4 digits job title Problems: * current classification: very useful for SES, but job titles with completely different exposure profiles into similar codes * old classification: not useful for modern jobs (e.g ICT is completely lacking) or jobs with strong development (eg order picker in warehouse) * do not underestimate inter-observer agreement !
How well is the classification of occupational job titles ? Study population with 37 occupational OJT (each at least 30 subjects) Self-reported physical load: often to always standing 23% sitting 50% walking 18% manual materials handling > 5 kg 14% manual materials handling > 25 kg 2.1% driving vehicles > 4 hrs 1.4%
How well is the classification of occupational job titles ? Study population (n > 3000) with 37 OJT (each at least 30 subjects) Physical load prevalence explained variance by OJT standing 23% 28% sitting 50% 38% walking 18% 18% mmh > 5 kg 14% 12% mmh > 25 kg 2.1% 7.8% driving > 4 hrs 1.4% 2.9%
How well is the classification of occupational job titles ? Study population with 17 occupational job groups, comparable OJT collated into groups (hierachical taxonomy, 5 digits into 3 digit group) Physical load explained variance by OJT by OTG standing 28% 14% sitting 38% 27% walking 18% 11% mmh > 5 kg 12% 8.1% mmh > 25 kg 7.8% 7.0% driving > 4 hrs 2.9% 1.3%
How well is the classification of occupational job titles ? Conclusion: - JEM performance partly determined by between-worker (=within- group) and between-job variance, post hoc evaluation needed to determine optimum grouping strategy - how specific should the OJT be ? 5 digit level requires more information, hampers comparability - explained variance very low with low prevalence of exposure (specificity is crucial factor)
Internal versus external JEM Internal JEM: - information on distribution of exposure within the study population - artificial reduction of between-worker variance (zero within job) - available exposure data usually limited, hence, less precision External JEM: - information on distribution of exposure from external population or based upon expert judgement - generalisibility to other populations ?
Some classical issues Job-exposure matrix: - level of detail of exposure characterisation (eg 'possible' (lenient) and 'probable' (strict)) - performance relative to exposure prevalence: * low prevalence, specificity is more important * high prevalence, sensitivity becomes more important - JEM vs individual assessment * JEM less recall bias or no bias at all (blinded) * classical vs Berkson's error (ind vs grouping) * optimum grouping strategy