Efficiency of autocoding programs for converting job descriptors into standard occupational classification (SOC)...

Area of Science:

• Occupational Health
• Data Science
• Epidemiology

Background:

• Existing datasets frequently lack crucial job exposure information.
• Standard Occupational Classification (SOC) codes are vital for linking occupational exposures to health outcomes using Job Exposure Matrices.
• Manual assignment of SOC codes is time-consuming and resource-intensive.

Purpose of the Study:

• To evaluate the effectiveness of two automated SOC coding programs.
• To assess the utility of autocoding in linking free-text occupation data to exposure information.
• To compare automated SOC coding with manual assignment methods.

Main Methods:

• Two existing cohorts' industry and occupation descriptions were input into two public SOC autocoding programs.
• SOC codes were also assigned manually by experienced coders for comparison.
• Assigned SOC codes were linked to exposure data from the Occupational Information Network (O*NET).

Main Results:

• Autocoding programs demonstrated modest agreement with manual SOC code assignment at the 6-digit level.
• A strong agreement was observed between autocoded and manually assigned SOC codes at the 2-digit level.
• Crucially, O*NET exposure values derived from autocoded SOC assignments showed high concordance with manually derived values.

Conclusions:

• Both evaluated SOC autocoding programs are valuable tools for assigning SOC codes.
• These automated methods facilitate the linkage of occupational exposures to health data containing free-text occupation descriptors.
• Autocoding offers a practical solution for overcoming limitations in existing datasets regarding job exposure data.