Somebody has to decide what killed you, and there is a manual for it

Every national statistic about what people die of rests on a decision somebody
made about an individual death certificate. A certificate may list several
conditions; one of them becomes the underlying cause of death, and the
others become contributing causes.

The National Center for Health Statistics produces the instruction manuals
that govern how that decision is made, used by **mortality medical coders,
nosologists** and others who process vital records — at both state vital
statistics programmes and the NCHS itself.

What the manuals cover

They are split by topic:

• Classification and coding — underlying and multiple causes of death,
  geographic variables, and industry and occupation
• Specifications for the U.S. standard certificates and reports
• Comprehensive descriptions of the computer edits the NCHS uses in
  processing natality and mortality data

Some are updated annually, including **Part 2 — Instructions for Classifying
Underlying and Multiple Causes of Death**, which is split further:

The part worth pausing on

Part 2c is a set of decision tables, which is to say the choice is
substantially automated and rule-governed rather than left to judgement. That
is the only way national mortality statistics can be comparable: if two coders
in two states reading the same certificate reach different answers, every
trend line built on top becomes a record of coding practice rather than of
death.

It is also why revisions to these manuals matter beyond the office. A change
in how the underlying cause is selected can move a national cause-of-death
statistic without anybody dying differently.

Selected manuals are also available from the CDC's FTP server in WordPerfect
and ASCII format — a detail that tells you how long this system has been
running.

*Source: National Center for Health Statistics, Centers for Disease Control
and Prevention.*