Estimated Comparability Ratios
Starting with 1999 deaths, important and profound changes have been made to how causes of death are classified and tabulated. These changes are the latest episode of periodic revisions to the International Statistical Classification of Diseases and Related Health Problems (ICD) as the understanding of diseases, their etiology and their classification changes. The ninth revision of the ICD (ICD-9) has been in use in the U.S. from 1979 through 1998. (ICDA-8 was used 1968-78, ICD-7 from 1958-67, and ICD-6 from 1949-57.) ICD-10, or tenth revision, currently is used for cause-of-death classification.
The World Health Organization (WHO) promulgates the International Classification of Diseases, which includes the classification (definitions) of causes of death and the mortality coding rules. The coding rules are used to determine the single underlying cause of death from the multiple causes of death that may have been reported on the death certificate by the certifying physician, coroner or medical examiner. Each state, as well as much of the world, uses the ICD to code and classify causes of death.
The changes from ICD-9 to ICD-10 are considerable, and include revision of categories, re-groupings of diseases, renamed causes of death, and modification of coding rules. There are roughly 4,000 codes in ICD-9, whereas there are approximately 8,000 codes in ICD-10. With name and definition changes arise issues of comparability. For example, heart disease is defined differently in ICD-9 from how the category "diseases of heart" is defined in ICD-10. Inasmuch as cause-of-death information using ICD-9 definitions and coding rules would result in heart disease as the underlying cause of death, application of ICD-10 definitions and coding rules would result in some of those deaths being attributed to other causes. Similarly, under ICD-9 some deaths may have been coded to other causes of death that under ICD-10 would be coded to diseases of heart. In general, the net effect between the two ICD revisions for specific causes of death may be 99 percent or more comparable, whereas others may be 70 percent or less.
Nota Bene: ICD revisions do not affect the total number of deaths or "deaths due to all causes," as this number is determined simply by the number of recorded death certificates, but they affect the size of the cause-of-death "slices" in the total-number-of-deaths pie.
To help in understanding the differences in coding and classification due to the change in ICD revisions at the national level, NCHS did a recent comparability study. In this study, an 80 percent national sample of 1996 death certificates were cause-of-death coded twice: once using ninth revision codes, classifications and rules and again using those for the tenth revision. A comparability ratio of 1.0000 can be interpreted as the same number of deaths were found under both revisions of the ICD. A comparability ratio less than 1 means fewer deaths for a particular cause were coded under ICD-10 compared with the comparable cause under ICD-9. A comparability ratio greater than 1 means more deaths for a particular cause were coded under ICD-10 than the comparable cause under ICD-9.
It should be noted that the comparability ratio represents a net difference in the number of deaths classified for a particular cause between the two revisions of the ICD. That is, a cause of death may include additional deaths from a different cause while losing deaths to a third cause. One such example is diseases of heart, which had additional deaths under ICD-10 from pneumonia (compared to ICD-9), but lost deaths due to cardiac arrest, an ill-defined condition in ICD-10 (but part of heart disease under ICD-9). Therefore, a ratio of 1.00 does not necessarily imply perfect comparability. It also is possible there are deviations from a cause-specific comparability ratio by demographic variables, such as age, sex, race and geographic region.
For more information about the national comparability study, please see the CDC publication: National Vital Statistics Reports, Volume 49, Number 2, "Comparability of Cause of Death Between ICD-9 and ICD-10: Preliminary Estimates," by Robert N. Anderson, et al. This publication can be viewed at the following Internet address: www.cdc.gov/nchs/data/nvsr/nvsr49/nvsr49_02.pdf. To view this PDF formatted file, you will need the Adobe Acrobat Reader software.