Analyzing Death Certificate Data

• CDC code set used for Census codes:
  • 2018 Industry
  • 2018 Occupation
• Data set documentation for industry and occupation groupings
  • Industry Appendix page 624
  • Occupation Appendix page 628

• CDC code set used for Census codes:
  • 2012 Industry
  • 2012 Occupation
• Data set documentation for industry and occupation groupings
  • Industry Appendix page 461
  • Occupation Appendix page 465

Learn more about these code sets.

Note: The International Classification of Diseases 10^th Revision (ICD–10) codes included in the data are ICD–10 (mortality), not clinical modification (ICD–10–CM) codes.

• Underlying Cause of Death: This is the disease or injury which leads directly or indirectly to death. It can also be the circumstances of an accident or violence that leads to death.
• Contributing Causes of Death: This is referred to as "Multiple Conditions" in the record layout document. Often, it is the combined effect of two or more conditions that result in death. These conditions may be unrelated, independent of each other, or causally related (one cause may lead to another).

NCHS provides two types of codes for contributing causes of death:

1. Record axis codes are the NCHS-edited version of the contributing cause data from the death certificates. These codes are edited for inconsistent information, combine conditions listed separately and standardize the diagnoses. They do not include any information regarding their placement on the death certificate.
2. Entity axis contributing cause codes are the conditions from the death certificate listed in the same order that they appear on the death certificate. These data are unedited; the number of diagnoses and level of detail on the certificate may vary widely from one certifier to the other, even for the same conditions.

NOMS uses record axis codes. Both record and entity axis codes have important uses in mortality research, but they each serve different purposes. We use record axis codes to standardize data across different sources so that we can make comparisons in patterns of deaths by industry and occupation.

Mortality data include individual ICD–10 codes as well as groupings of ICD–10 underlying cause of death codes. For description of groupings and codes included, see 39-cause recode, 113-cause recode, and 358-cause recode in the file documentation and the Instruction Manual Part 9: "ICD–10 Cause-of-Death Lists for Tabulating Mortality Statistics."

See the NVSS Public Use File Documentation to learn about other available variables.

1. This SAS program [TXT – 11 KB] can be edited and then used to read the data into SAS. Read the file documentation and edit program as needed for any changes to variables or variable location.
2. This format program (2012 [TXT – 66 KB]; 2018 [TXT – 72 KB]) can be used with the read-in program to assign labels and formats.
3. This separate program [TXT – 770 KB] includes formats for individual ICD–10 codes.

For the two different filename types:

1. For .txt filename extension: For the text data, this R program [TXT – 4 KB] can be edited and then used to read the data into R.
2. For .dusmcpub filename extension (2020 data file): If you must read in the .dusmcpub file, be aware that you may need to run R overnight to fully import the data. For more information on reading this filename type into R, see the following resource.

Tip: When using R with the .dusmcpub filename type, it is best to save the data as a .csv file. This will prevent overly long loading times. If you have SAS, read in the data using this SAS program [TXT – 11 KB] and export the data as a .csv file that can be more easily read into R.

Use the following to group Census Industry and Census Occupation codes into detailed NHIS industry and occupation recodes.

SAS (2012 | 2018) R codes (2012 | 2018)

• 544 individual CDC Census Occupation (2012) codes → 96 NHIS occupation groupings.
• 272 individual CDC Census Industry (2012) codes → 80 NHIS industry groupings.

Simple recodes of industry (22 groupings) and occupation (25 groupings) variables are included in the public use data set.

If you are an experienced SAS user, use the following SAS code [TXT – 29 KB] to compute adjusted PMRs.