Steps for use: Job Exposure Matrix for exposures associated with occupational asthma

Before following these steps, please make sure you have read the background section [Why use this tool].

Step A: Check job coding:

  1. Jobs coded to major group codes only should be recoded or excluded from analysis
  2. Many of the submajor and minor group codes require recoding – see the comments column.
  3. Jobs may be classified under more than one exposure category. The ‘uncertain’ category indicates job codes that are considered to be only poorly classified by this system.

Step B: Initial merging of the job exposure matrix to your dataset:

  1. Start with a dataset from your own study that is set up to have one row per job (so if a person has more than one job, she or he will have more than one row in the dataset). The dataset should include (as a minimum) the person id variable, the ISCO job code, and one or more variables to identify which job this is for the person (eg. start and stop dates or job number). If possible, the data set should include the TEXT that describes the job and industry (eg. job title/ industry title/ and brief descriptions).
  2. Merge this job exposure matrix to your data set (merging on ISCO code) using whatever computerized merging system you have available. If you have a sas data set, you can modify the SAS code provided here to do this step.
  3. You should now have a data set in which each row identifies a job held by your study subject (with the same original variables you had in your own dataset) PLUS 29 new variables. The new variables include 2 ‘CHECK’ variables, 18 ASTHMA HIGH RISK exposure variables (including 1 RADS exposure risk variable), 4 ASTHMA LOW RISK expsoure variables, 1 ASTHMA NO OR VERY LOW RISK variable, 1 variable to identify jobs with large uncertainty in the exposure estimates, 1 variable with comments, and 1 variable for your own use if you want to flag anything.

Step C: First level of 'verification' - checking ISCO job coding:

  1. Sort the data set by the variable ‘check ISCO’. Review all the ISCO codes for the jobs so identified, and make changes to ISCO codes as instructed in the ‘comments’ column.
  2. Re-check all remaining job title codes (ie. ISCO codes) by sorting the resulting data set by the ISCO code variable and examine your job title / industry TEXT – not the standard ISCO text – rather the text from your original data set (for each job). The purpose of this step is find out if you have any remaining ISCO coding errors in your original data set. If you identify errors in the ISCO coding, correct them now.
  3. Save a new version of your dataset. Remerge this new dataset (ie. with corrected ISCO codes) with the job exposure matrix again (ie. go back to step B and repeat it).

Step D: Second level of 'verification' - checking / correcting the exposure codes for SELECTED jobs:

  1. Sort the dataset by the ‘check exposures’ variable and identify all jobs coded as ‘1’ for this variable. Look at the instructions / suggestions listed under the variable ‘comments’ and consider whether or not it is necessary to reassign one or more of the exposure variable codes for these jobs.
  2. The 4 ‘low asthma risk exposure’ variables are included so you can keep track of other exposures that may contribute to airflow obstruction if you wish. You may find it useful to use these variables (especially the one labelled ‘low asthmagen’) instead of changing one of the high risk asthma exposure variable codes if you are not certain you should make a change.

IMPORTANT: During this step, only change those codes as directed by the comments section. If you are not sure what to do, LEAVE THE CODING AS IS.