Alan Rozanski

Who they are

Rozanski is a cardiologist whose research concerns the relationship between psychological factors and cardiovascular disease. His 2019 paper in JAMA Network Open is the field's standard pooled estimate and it is open access, which is part of why its figures travel so easily.

The method is conventional and well reported. PubMed, Scopus and PsycINFO were searched from inception to 2 July 2019, the MOOSE reporting guideline was followed, data were screened and extracted independently by two investigators, and the pooling used a Hartung-Knapp-Sidik-Jonkman random-effects model. Fifteen studies qualified, ten reporting cardiovascular events and nine all-cause mortality, with a mean follow-up of 13.8 years and a range from two to forty. Restricted to studies using the revised Life Orientation Test, the estimates were 0.71 for events and 0.87 for mortality.

The paper's most valuable section is its limitations, and it is more candid than most. Although each included study adjusted for important covariates, the covariates varied widely from study to study, so the authors could not systematically assess the independent effects of individual clinical variables including smoking, diabetes and hypertension, and they say this variability may help to account for the considerable heterogeneity found among studies. An I-squared of 87.4 per cent means the included studies are not estimating one quantity. A pooled relative risk across such studies is a summary of a disagreement rather than a measurement, and the authors' own text is what licenses saying so.

Two further decisions in the paper shape the whole field's evidence base. Studies of patients with cancer were excluded by design, in the paper's phrase to avoid confounding secondary to terminal sickness. That means the pooled estimate everybody quotes for optimism and mortality contains no cancer patients, and the cancer question has had to be asked separately, with different and largely null answers at the positive pole. And the mechanisms section lists associations with inflammation, haemostasis and endothelial function, metabolic function, telomerase activity and telomere length, ambulatory blood pressure and hypothalamic-pituitary-adrenocortical function, while stating plainly that study of individual mediators remains sparse compared with better-established psychosocial mediators such as depression.

His team also raises the asymmetry that our article is built on, before the evidence for it was assembled: they write that prospective study is needed to evaluate whether pessimism is a stronger contributor to pathophysiologic dysfunction, as some studies suggest, than optimism is for providing positive physiologic buffering. And they state the question that stands between this literature and any advice a reader could act on, which is whether optimism enhanced or induced through directed prevention or intervention strategies has health benefits similar to optimism that is naturally occurring.

Key papers

Rozanski is a clinician and researcher whose relevant output here is journal articles rather than books.

Rozanski, A., Bavishi, C., Kubzansky, L. D., & Cohen, R. (2019). "Association of Optimism With Cardiovascular Events and All-Cause Mortality: A Systematic Review and Meta-analysis." JAMA Network Open 2(9), e1912200, doi:10.1001/jamanetworkopen.2019.12200. Read in full for this article, open access, including the mechanisms, clinical-significance and limitations sections.

Key ideas

Pooled across 15 studies and 229,391 participants, optimism is associated with a relative risk of 0.65 for cardiovascular events and 0.86 for all-cause mortality.

Heterogeneity is high, 87.4 per cent for events and 73.2 per cent for mortality, and his team attributes it partly to variability in covariate adjustment between studies.

Studies of cancer patients were excluded by design, so the pooled estimate contains none.

Individual physiological mediators are sparsely studied compared with depression, and the list of candidate pathways should be read at that strength.

Whether optimism raised by an intervention carries the same benefits as naturally occurring optimism is unknown, and his paper says so in print.

Pessimism may be the stronger contributor to pathophysiological dysfunction, which his team names as a question for prospective study.

---