in effect for deaths that occurred in 1960–1967, 1968–1978, and 1979–1998, respectively; the differences among them are fairly subtle. Although ICD-10, which went into effect for coding causes of deaths that occurred from 1999 on, appears radically different from the earlier versions, it corresponds in large part to basically the same disease entities (see Table B-2). To date, most published epidemiologic studies have been related to health outcomes that occurred and were encoded before ICD-10 went into effect.
Since 1983, the National Institute for Occupational Safety and Health (NIOSH) has maintained software for generating standardized expectations, as derived from US mortality data assembled by the National Center for Health Statistics, for ICD-encoded mortality data sets. An article by Robinson et al. (2006) discusses revisions to that standard software to incorporate deaths coded according to ICD-10 and includes conversions and equivalencies between ICD-7, -8, -9, and -10 for 119 exhaustive categories for cause of death. Codes for malignant neoplasms span the ICD-9 range 140.0–208.9, NIOSH’s major categories 02–10, or NIOSH’s more specific minor categories 004–040.
The NIOSH death codes for neoplasms provide comprehensive scaffolding for organizing the committee’s reviews and conclusions by cancer type that is somewhat simpler than ICD classifications, but maps completely to the ICD system as it has evolved. Because the NIOSH system has been used to mediate analysis of many sets of cohort data, its groupings correspond quite closely with the published research findings available for VAO review. In general cohort studies, one is unlikely to encounter results on more specific groupings than NIOSH’s minor categories.
As discussed in Chapter 2, this committee has not framed its conclusions strictly in terms of ICD codes, but the ICD system has been a valuable tool for the work of VAO committees. There can be coding errors on hospital records or death certificates, but when researchers present their results labeled with ICD codes, there can be little ambiguity about what they intended. When their most definitive indication is something like “respiratory cancers,” however, there can be uncertainty about where the evidence should be considered. In such cases, the committee has done its best to follow the hierarchy laid out in Table B-1.
As indicated above, many of the studies reviewed by the committee use or were written at a time when ICD-9 was in place. Accordingly, ICD references in this report use that scheme. ICD-10 began to be implemented in the United States in 1999. It differs from ICD-9 in level of detail (about 8,000 categories versus about 5,000 in ICD-9) and nomenclature (alphanumeric versus the numeric codes of ICD-9); additions and modifications were also made with regard to some coding rules and the rules for selecting an underlying cause of death (Anderson et al., 2001). Table B-2 lists the ICD-9 and ICD-10 codes for the various forms of malignant neoplasm addressed in this report. In situ neoplasms, benign neoplasms, neoplasms of uncertain behavior, and neoplasms of unspecified behavior have separate codes in both schemes.