Data integration of National Dose Registry and survey data using multivariate imputation by chained equations

Area of Science:

• Occupational health
• Radiation safety
• Data science

Background:

• Diagnostic medical radiation workers were enrolled in the National Dose Registry (NDR) from 1996 to 2011.
• Linking NDR data with mortality and cancer registries provides comprehensive health information.
• A survey on occupational radiation practices was conducted among these workers.

Purpose of the Study:

• To integrate data from the National Dose Registry (NDR) and a survey on occupational radiation practices.
• To assess the effectiveness of data integration for obtaining worker information without new surveys.

Main Methods:

• Data integration was performed using the multivariate imputation by chained equations (MICE) algorithm.
• The study integrated NDR data with self-reported occupational radiation practices.
• Results were analyzed comparing observed and imputed data, stratified by sex and job type.

Main Results:

• Differences were observed between imputed and actual data for interventional therapy frequency in nurses, influenced by medical facility distribution.
• Marital status, pregnancy status, and smoking history showed discrepancies between observed and imputed data due to variations in birth year and drinking frequency distributions.
• Imputed data indicated a higher percentage of smoking experience compared to observed data, potentially due to underreporting.

Conclusions:

• Data integration of NDR and survey data is a viable method for obtaining occupational radiation practices information.
• This approach significantly reduces the need for additional surveys, saving time and costs.
• The study highlights the utility of data integration in epidemiological research involving radiation workers.