Lewina O. Lee

Who they are

Lee works at the National Center for Posttraumatic Stress Disorder at VA Boston Healthcare System and in the Department of Psychiatry at Boston University School of Medicine, where her research concerns how psychological characteristics relate to physical health and ageing. Her 2019 paper in the Proceedings of the National Academy of Sciences is the one the field, and the press, quote.

The design is worth stating carefully, because the detail is what makes the result defensible. Optimism was measured in the women in 2004 with the revised Life Orientation Test, and deaths were tracked to 1 June 2014, with a mean baseline age of 69.9. The 1,400 women who died within the first two years were excluded, in the paper's own words to mitigate the possibility that imminent health decline affected their reports of optimism. In the men, optimism was measured in 1986 with the Revised Optimism-Pessimism Scale, derived from 263 items of the Minnesota Multiphasic Personality Inventory-2, with mortality tracked to the end of 2016; 71 per cent of the men died during follow-up. The two cohorts therefore used two different instruments, correlated only moderately in prior work, which the paper notes, so this is a replication across populations rather than across one measure.

The adjustment is unusually thorough for this literature. For the nurses, the models carry baseline age, race, marital status, the participant's own education, her husband's education and her father's occupation when she was sixteen, then high cholesterol, hypertension, type 2 diabetes, myocardial infarction, stroke, cancer and depression, then smoking, physical activity, alcohol, diet quality on the Alternative Healthy Eating Index, attendance at a screening physical examination and body mass index. Adjusting for childhood social class and marital socio-economic position is more than most psychology papers manage, and it deserves saying.

The sensitivity analyses run in the same honest direction. Excluding everyone with a major chronic disease at baseline left the findings comparable, with fewer reaching statistical significance on the reduced samples. Adding social integration, measured on the Berkman-Syme Social Network Index, attenuated the estimates somewhat while leaving them significant in both cohorts. The paper also supplies its own calibration: in the same model, never having been diagnosed with type 2 diabetes was associated with a 17.0 per cent longer life span for the women, and never having had a heart attack with 18.0 per cent, so optimism's 14.9 per cent sits just below both.

Her team's own limitations are stated plainly in the paper. Both cohorts were mostly white and of higher socio-economic status than the general population, so the findings may not generalise to minority or more disadvantaged groups. Reverse causation remains possible, and their answer to it is the exclusion of early deaths and the chronic-disease sensitivity analysis rather than a claim to have ruled it out. Recorded for completeness and not for insinuation, the paper declares that a co-author has worked as a consultant with AARP and United Health Group.

Key papers

Lee is an academic epidemiologist and psychologist whose output is journal articles rather than books, so this section lists papers rather than a bibliography of titles.

Lee, L. O., James, P., Zevon, E. S., Kim, E. S., Trudel-Fitzgerald, C., Spiro, A. III, Grodstein, F., & Kubzansky, L. D. (2019). "Optimism is associated with exceptional longevity in 2 epidemiologic cohorts of men and women." Proceedings of the National Academy of Sciences 116(37), 18357-18362, doi:10.1073/pnas.1900712116. The source of the 11 to 15 per cent longer life span figures and of the odds ratios for reaching 85. Read in full for this article, including both tables, the Methods, the Discussion and the reference list. Author order confirmed independently.

Key ideas

Optimism, measured once on a short questionnaire, is associated with a longer life span in two large cohorts, at 14.9 per cent for women and 10.9 per cent for men after adjustment for demographics, baseline conditions and depression.

Adding health behaviours cuts the women's figure to 8.7 per cent and barely moves the men's, from 10.9 to 9.8. The sex difference is large and unexplained.

Health behaviours may be confounders of this association and may also lie on its causal pathway, so both estimates are defensible and observational data cannot decide between them. This is the sentence the whole argument about the figure turns on.

Optimism is roughly 25 per cent heritable, shaped by social structural factors, and can be learned; the five-year correlation of the men's score was 0.87, which is what makes it trait-like.

Both circulated figures come from the model before behaviour adjustment, and the behaviour-adjusted odds ratios for reaching 85, 1.2 and 1.6, have lower bounds at or touching 1.0.

---