Longevity disparities in multiethnic Hawaii: An exemplary analysis with a small, significant error
This is a brief review of:
In the cited study, Park et al. report on the life expectancy in Hawaii from 1950 to 2000, stratifying their analyses by ethnic groups. Ethnicities considered included Caucasian, Chinese, Filipino, Hawaiian, Japanese, Korean, and Samoan. In Table 1 the authors report life expectancies from birth for the various ethnic groups for 1950 to 2000 in ten year increments. Life expectancies increased steadily over the years for every ethnic group, the increases from 1950 to 2000 ranging from 14% (Caucasian, Japanese) to 24% (Chinese). In the year 2000, life expectancies from birth for the various ethnic groups varied rather dramatically, from a low of 72.8 (Somoan) to a high of 86.1 (Chinese). The authors provide interesting and informative discussion regarding possible reasons for the differences. They note that biological factors are important, but they attribute the greater part of the variation to behavioral and social factors, a hypothesis they argue is supported by national and local data sources.
In addition to reporting life expectancies from birth for the various ethnic groups and decades, Park et al. also reported age-specific mortality rates (in broad age groups) in Table 2 of their study. With few exceptions the age- and ethnicity-specific mortality rates reported in Table 2 are consistent with the rank order of life expectancies reported in Table 1. For example, the mortality rates in Table 2 for Chinese residents of Hawaii are generally lowest, corresponding to their life expectancies in Table 1 being highest.
However, a small but significant error apparently made its way into Table 2 in this study. While the error does not affect the validity of the comparisons made in the previous paragraph (for the rank order of the rates is likely correct), the magnitudes of the mortality rates are apparently erroneous. The heading of Table 2 reads: Life table death rates per 1,000 population by ethnicity for broad age groups, Hawaii, 2000. It seems, however, that the actual numbers reported are death rates per 100 person-years, rather than 1,000.
According to the latest life tables published by the NCHS, the average mortality rate for US white males from age 0 to 14 is roughly 0.7 per thousand. According to Table 2 of Park et al. Caucasian males in Hawaii have a mortality rate of 0.07 per 1,000, making the rates from Park et al. lower by a factor of 10. If we use the rates reported in Table 2 in a life table, we can estimate life expectancies for the various ethnic groups and compare them to the life expectancies listed in Table 1 of Park et al. In the year 2000, for example, Caucasians males in Hawaii are reported to have a life expectancy from birth of 76.7 years (Table 1 of Park et al.), whereas, if we use the age-specific mortality rates (Table 2 of Park et al.) in a life table, the resulting life expectancy for Caucasian males is greater than 140 years. Assuming instead that the rates in Table 2 are actually per 100, the resulting life expectancy from birth for Caucasian males is 76.7 years, matching perfectly the corresponding figure in Table 1. Both of these figures are comparable to the corresponding US life expectancy for white males in 2000 reported by the NCHS. It seems likely then that the rates in Table 2 are indeed per 100 population rather than per 1,000 population.
We should add that shortly before Christmas 2012 we wrote to the corresponding author of the cited study, Dr Chai Bin Park of the University of Hawaii's John A. Burns School of Medicine to ask his thoughts about this (likely typographical) error in his Table 2. The day after Christmas we received a gracious reply from Dr Park's daughter, Dr S. Lani Park, who relayed that her father had taken ill and was unable to review his work at that time.
Yesterday we sent another e-mail to Dr Lani Park inquiring about her father. We sincerely hope he is well, and in spite of what we believe was a small error in one table in this study we commend him and his colleagues for their interesting and informative work.