Hayami K. Koga

Who they are

Koga is an epidemiologist at the Harvard T. H. Chan School of Public Health. Her 2022 paper in the Journal of the American Geriatrics Society is the replication that the earlier nurses-and-veterans result needed, and it is a better-designed study in three specific ways.

The sample is large and more diverse than its predecessors: 131,666 non-Hispanic White, 14,261 Black, 6,287 Hispanic or Latina, 4,144 Asian, 693 American Indian or Alaskan Native and 1,810 other participants, enrolled between 1993 and 1998 from 40 clinical centres and followed to 1 March 2019, by which point 64,301 of them, 40.3 per cent, had died. Optimism was measured with the six-item revised Life Orientation Test, alpha 0.75. Her headline figure, adjusted for demographics including income, occupation, health insurance and partner's education, plus chronic conditions and depression, is a 5.4 per cent longer life span in the most optimistic quarter, on an interval from 4.5 to 6.4, falling to 4.59 per cent once lifestyle enters.

The mediation analysis is the second improvement. Rather than inferring the behavioural contribution from how far an estimate moves between models, her team ran a formal counterfactual decomposition, allowing for exposure-mediator interaction because they found one, and reported that lifestyle factors mediated 24 per cent of the association. Their lifestyle composite, scored from zero to five on diet quality, physical activity, body mass index, smoking and alcohol, was measured three years after the optimism assessment, which fixes the temporal order instead of assuming it. Their own conclusion is that the contribution of lifestyle to these associations was modest.

The third is the subscale split. Taking the two three-item halves of the questionnaire separately, the most pessimistic quarter showed a 5.2 per cent shorter life span, 4.3 to 6.1, and the most optimistic quarter a 2.5 per cent longer one, 1.6 to 3.4. The pessimism half is more than twice the size of the optimism half, in her team's words slightly larger in magnitude, and the finding lines up with the consortium reanalysis and with the genetic evidence.

Her paper is also where the circulated "optimists live four years longer" figure comes from, and it arrives properly labelled. Taking United States female life expectancy in 2018 of 81.2 years, the paper says a 5.4 per cent increase can be extrapolated as adding 4.4 years of life on average, then compares that to the 0.4 to 4.2 years reported for regular exercise. That is transparent arithmetic described as arithmetic. What happens to the sentence afterwards is not her doing.

Her limitations are her own and they travel with every use of the paper. The participants were better educated than the average United States population, and 53 per cent of those eligible reached 90, which suggests they were healthier too, so caution is needed in generalising to less long-lived populations. Unmeasured confounding by factors including heritability and childhood environment, parental warmth and financial stability, may be possible.

Key papers

Koga is a working epidemiologist whose output is journal articles, so this section lists papers rather than books.

Koga, H. K., Trudel-Fitzgerald, C., Lee, L. O., James, P., Kroenke, C., Garcia, L., Shadyab, A. H., Salmoirago-Blotcher, E., Manson, J. E., Grodstein, F., & Kubzansky, L. D. (2022). "Optimism, lifestyle, and longevity in a racially diverse cohort of women." Journal of the American Geriatrics Society 70(10), 2793-2804, doi:10.1111/jgs.17897. The accepted manuscript, including all three tables, was read in full for this article.

Boehm, J. K., Chen, Y., Koga, H., Mathur, M. B., Vie, L. L., & Kubzansky, L. D. (2018). "Is Optimism Associated With Healthier Cardiovascular-Related Behavior? Meta-Analyses of 3 Health Behaviors." Circulation Research 122(8), 1119-1134, doi:10.1161/CIRCRESAHA.117.310828. Koga is third author. Abstract read for this article; full text not read.

Key ideas

The optimism and longevity association replicates in a much larger and more diverse cohort, at 5.4 per cent longer life span and a risk ratio of 1.10 for reaching 90.

Lifestyle mediates 24 per cent of it on formal causal mediation, with the mediator measured three years after the exposure. That is a quantity in place of a choice between two models.

Split the questionnaire and the pessimistic half does more than twice the work of the optimistic half.

Her Table 1 is the clearest statement of the confounding problem in the whole literature: across quartiles of the same questionnaire, depression runs at 22 per cent against 4, below-secondary education at 22 against 9, and family income under twenty thousand dollars at 23 against 10.

Two of four racial and ethnic subgroup intervals cross zero, and the risk ratios for reaching 90 in Asian and in Hispanic or Latina women are null. Her team reports no significant interaction and attributes the imprecision to sample size, which is reasonable and is not itself a result.

The four-years figure is her team's own extrapolation from a percentage, labelled as one in a single sentence of the discussion.

---