How to Read a (Good) Research Abstract

Just how risky is that breakfast?  I like this abstract published in the Journal of the American Medical Association.  It describes a study of the effect of egg intake on mortality risk over the course of 17.5 years of follow-up.   

If you know how to read and interpret the abstract's key points, you can translate them into very meaningful, concrete terms.  Here's how to read the abstract's key sentence:

"Each additional half an egg consumed per day was significantly associated with higher risk of [...]  mortality (adjusted HR, 1.08 [95% CI, 1.04-1.11]; adjusted ARD, 1.93% [95% CI, 1.10%-2.76%])."

1. Translation:  For each extra half an egg eaten per day, mortality from all causes was estimated as...

• 8% higher in relative terms (multiplied), which meant
• 1.93% higher in absolute terms (added).

2. Three definitions to help decipher that key sentence:

• HR = hazard ratio:  the multiplier for the risk of mortality linked with half an egg extra per day, while controlling for other factors.
• ARD = absolute risk difference:  the added extra risk, in percentage points.
• CI = confidence interval:  a reasonable range for each result. 

3. The findings in concrete terms:

• Background fact:  21% of all study participants died of whatever cause during the 17.5 years.
• For those who ate half an egg above the average per day, mortality risk was estimated as 21% times the HR of 1.08, or 23%. 
• This works out to 21% plus 1.93 percentage points, which also is 23%.
• Using the CI for the ARD, for these extra egg-eating people we arrive at a range for mortality risk between 22% and 24%.  The CI tells us that we add at least 1.10% of risk.  With 95% confidence, then, the added risk is greater than zero; thus at the .05 level the finding is statistically significant.

4. Does it matter?
Armed with this information we can each make our own informed decision as to whether the findings have not just statistical significance but practical significance.  What do you think?  Would you skip an extra omelette per week to avoid a 2-percentage-point increase in 17.5-year mortality?

​Contact:  [email protected]