The findings, published today in Nature Communications, show that good access to food at the pre-pubescent age of nine to 12 means their grandsons – but not their granddaughters – die on average earlier, especially from cancer.

Economics professor Gerard van den Berg, from the University of Bristol, and collaborating researchers from the Universities of Stockholm and Bonn looked at data from a large, three-generational study.

It expands on previous work and supports the hypothesis that experiences and environmental exposures can change the way a person's DNA works and influence health outcomes in subsequent generations – known as transgenerational epigenetic inheritance.

Researchers collected information on the yields of harvests in Sweden for the regions included in the Uppsala Birth Cohort Multigeneration Study from 1874 to 1910.

This data was used to estimate how difficult it was for the 9,039 grandparents in the cohort to access food during their so-called 'slow growth period' from the age of nine to 12.

Using data on mortality from 1961 to 2015 for their 11,561 grandchildren, the authors found that paternal grandfathers who had access to unusually high harvest yields had grandsons who died earlier than otherwise. Interestingly, food access among maternal grandparents does not seem to play a similar role.

Cancer deaths appear to be the driver for the 'highly significant' result, as no correlation was identified with cardiovascular and diabetes mortality.

Professor Gerard Van den Berg, an expert on health economics from the School of Economics, Finance and Management at the University of Bristol, said: "Our results suggest the existence of a transgenerational pathway along the male family line, triggered by environmental factors.

"While we have to be cautious about the mechanism of inheritance, the implications are substantial in terms of how we understand evolution, and they are important for the design of economic policies to support children."

Further studies showing a direct effect on epigenetic markers in the germline and transmission of such markers over three generations would be required to unambiguously establish the pathway from a grandfather's access to food and a grandson's mortality risk.