It is really a good example of how to lie with statistics and a headline.
"The overall, male, and female death rates for the 10-year period were 0.75 (95% confidence interval [CI], 0.38-1.13), 0.98 (95% CI, 0.48-1.36), and 0.41 (95% CI, 0.21-0.79) deaths per 100,000 finishers, respectively. There was no change in the death rate during this time period for overall, male, or female groups (P = .860, .533, and .238, respectively). The median age among deaths was 41.5 years (interquartile range, 25.5 years). Fifty percent (14/28) of deaths occurred in participants less than 45 years old. Myocardial infarction/atherosclerotic heart disease caused 93% (13/14) of deaths in those 45 years and older. A variety of conditions caused death in younger racers, the most common being cardiac arrest not otherwise specified (21%, n = 3)."
Real headline would be:
"Study finds barely anyone that finishes a marathon dies but if they do, the median age is 41.5"