Honolulu Heart Study

Background/Question:

Around 1950, Tavia Gordon reported that the overall mortality rates for men in the United States and Japan were similar, yet the incidence of coronary heart disease (CHD) and cerebrovascular accidents was drastically lower in Japan. [1] This observation initiated three cooperative cohort studies, one of which was the Honolulu Heart Study. The purpose of the study was to determine whether there was a difference in CHD incidence and mortality between Japanese living in Japan and individuals of Japanese ancestry living in Hawaii.

Methods:

All participants were “non-institutionalized men of Japanese ancestry, born 1900-1919, now resident on the island of Oahu.” [2] To locate subjects, a clerk inspected 165,000 selective service registry cards from World War II, looking for birthdates between 1900 and 1919, among those with an apparent Japanese surname or notation of Japanese national origin. [2] Of the 22,892 cards that met these criteria, 12,417 lived in Oahu and had a current mailing address. Of the original mailing, 1,269 questionnaires were returned to sender and 1,270 men declined to participate. Of the respondents, 1,692 refused examination and 180 died before the study commenced, leaving 8,006 participants. [2]

The interview and physical examination phase began in October 1965. The mailed questionnaire obtained baseline demographic and medical information. The interview ascertained family and personal history of illness, sociological history, smoking status and physical activity level. As part of a complete physical examination, ECG and urinalysis were performed, and measurements of weight, height, skinfold thickness, blood pressure and serum cholesterol were taken. [3] Surveillance was conducted in cooperation with Oahu hospitals, which recorded the “diagnosis of any type of heart disease, CVA, or pulmonary embolus” and “abnormal electrocardiograms.” [3] Participants were also periodically mailed questionnaires on illnesses “suggestive of cerebrovascular disease or CHD.” [3] Mortality was measured by daily reviews of death certificates filed at the Hawaii State Health Department and the obituary section of local newspapers. [2]

Results:

The first four years of follow-up in the Honolulu Heart Study revealed “a rapid fall in mortality rates among non-respondents to a level equivalent to that of the respondents by the third year of observation; a consistently higher mortality among the non-ascertained part of the Japanese male population on Oahu[…] with every likelihood that this difference will persist, since this group contains a disproportionately large number of men who are in chronic care institutions, were never married, or are divorced, or separated from their wives; a clearly lower age specific mortality among the Issei (born in Japan) (age 60-64) as compared to Nisei (born in Hawaii) where there are enough of them to be compared, with more than half the difference being accounted for by coronary thrombosis.” [2] CHD mortality for Hawaiian men in all age groups occurred in about 5 men per 1000. [4] CHD incidence increased with age, with a rate of 1 case per 1000 men aged 50-54, and 4 cases per 1000 men aged 60-64. [4]. Over 23 years of follow-up, 864 cases of CHD were documented in the cohort, with 384 being fatal. [5] Multiple logistic regression was performed to analyze risk factors. This analysis revealed that age, systolic blood pressure, serum cholesterol, serum glucose, cigarette smoking and alcohol consumption were highly significant risk factors (p< .0001). [6] A number of studies on diabetes and CHD incidence and mortality have been performed on this cohort, as well, indicating that diabetes is a risk factor for CHD, with a relative risk of 2.82 (95% CI 2.27-3.50) for participants with diabetes versus a relative risk of 1.18 (95% CI: 1.01-1.38) for non-diabetics with high normal glucose tolerance. [5]

Conclusions/Discussion:

The authors acknowledged the non-representative nature of the study population, as it came from a non-random sample. [2] The method of selecting cases, however, allowed excellent surveillance of both respondents and non-respondents. The investigators were able “to validate our estimate of the number, age and birthplace of the non-respondents,” thus providing the means to later ascertain cause of death in non-respondents. [2] Follow-up through 23 years was virtually complete. [5] However, the analysis of “pre-existing non-genetic variables as possible factors in ASHD” was only performed on the respondents, and comparisons such as “the men of Japanese ancestry on Oahu have a higher prevalence of hypertension than do those in Hiroshima” were unable to be made. [2] (HB)