Twenty-year trends in mortality from chronic obstructive pulmonary disease: The Honolulu Heart Program

There is evidence of a rising trend in COPD mortality. Whether this is due to changes in coding or diagnostic practices, increases in disease incidence or severity, or other causes is unknown. The Honolulu Heart Program (HHP) has followed a cohort of 11,136 Japanese-American men, 45 to 65 yr of age at onset, from 1965 to 1984. Following a fixed protocol, study physicians assigned cause of death after review of hospital records, pathology and autopsy reports, the death certificate, and, in doubtful cases, after interview with family and personal physician. The eighth revision of the International Classification of Diseases was used throughout. During 20 yr of follow-up, 2,624 men died, 113 from COPD by HHP coding and 105 from COPD by State Health Department (HD) coding. There was no change in age-adjusted or age-specific COPD mortality rates from 1965 to 1984 by HHP coding (tests for trend, p > 0.05). In contrast, when HD coding was used there was a significant decline in COPD mortality over the same time period. The frequency with which COPD was found on the death certificate decreased significantly for diagnoses listed in Section 1 and increased for those in Section 2. Current smokers showed increasing trends and past smokers decreasing trends in COPD mortality. When HHP and HD codings were compared, there was agreement in only 46% (69 of 149) of the cases. Agreement was greater in 1965-1974 than in 1975-1984. This study found no evidence of increasing mortality rates from 1965 to 1984 in this cohort of Japanese-American men. Current practices in death certification indicated decreasing trends in COPD mortality and led to almost half of the COPD deaths being assigned to other causes. On a nationwide level this could contribute to significant inaccuracies in the rates and trends in COPD mortality.