Multiple cause-of-death data as a tool for detecting artificial trends in the underlying cause statistics: a...

Area of Science:

• Public Health
• Medical Statistics
• Epidemiology

Background:

• Cause-of-death statistics are crucial for public health surveillance.
• Trends in these statistics can be influenced by coding practices, not just disease prevalence.
• Understanding coding changes is essential for accurate interpretation of mortality data.

Purpose of the Study:

• To identify trends in underlying cause-of-death statistics attributable to coder selection and coding changes.
• To determine shifts in coder registration principles explaining observed statistical trends.

Main Methods:

• Analysis of 31 Basic Tabulation List categories from the Swedish national cause-of-death register (1970-1988).
• Estimation of the underlying cause ratio (coder tendency to register a condition as the underlying cause).
• Comparison of underlying cause rate trends with multiple cause rate trends and coder interpretation practice.

Main Results:

• Trends in thirteen categories (fourteen trends) were explained by changes in coder interpretation practices.
• Five upward, four insignificant, and five downward trends were identified due to coder interpretation.
• Three categories showed trends explained by new explicit International Classification of Diseases, 9th Revision (ICD-9) rules.

Conclusions:

• Coder interpretation practices significantly influenced cause-of-death statistics trends from 1970 to 1988.
• Changes in coding rules and interpretation led to observable shifts in underlying cause registration.
• Accurate interpretation of mortality data requires consideration of evolving coding practices.