Outcomes in Journal Articles

Outcomes can be expressed in a multitude of different ways that can be initially confusing for learners. Odds, Risks and Rates have different definitions but are sometimes used interchangeably in the literature.

Risks are the most commonly used and the easiest to understand. Risk represents the number of times the outcome of interest happened in each group we are analyzing. Risks are usually expressed either in decimals or in percentage. Even though the word "Risk" has a negative connotation, in EBM, it only means a probability. It could be a good or a bad outcome. We can be interested in a Risk of Survival, a Risk of Recovery, a Risk of Acute Kidney Injury or a Risk of Mortality.

i.e.: In group A we enrolled 80 patients. At the end of the study, 60 were alive and 20 passed away. We want to calculate the Mortality Risk.

Mortality Risk = 20 (patients that died) / 80 (Total number of patients).

Mortality Risk = 20/80 = 0.25.

Mortality Risk (in Percentage) = 100 * 0.25 = 25%.

Now that we have a way to calculate risks (probabilities), we can use them to compare what happened in the different groups of patients. The only ways to compare groups is to do subtractions or divisions. These lead to Risk Differences (subtractions) and Risk Ratios (divisions).

(Absolute) Risk Differences are the easiest to understand. If the experimental group has a mortality rate of 2% and the control group has a mortality rate of 5%, the difference is a 3% reduction in risk or rate of mortality [2%-5%=(-3%)]. Experimental always goes first by definition. A negative number means there is a reduction in risk and a positive number means an increase in risk compared to control (or baseline)

Risk Ratios are also fairly simple; the stumbling point is that there are multiple ways of expressing the data. Some articles might talk about Hazard Ratios, Risks Ratios (AKA Relative Risk) or Odds Ratios. By definition, Experimental risk should be the numerator and control risk should be the denominator.

In the next blogs we will tease apart the differences in the above ratios. What matters for now is that independently of what the metric is called, the interpretation follows some simple rules:

A ratio is always just a ratio....

A ratio of 1 means no difference in what we are comparing.

A ratio higher than 1 means the experimental groups outcome is higher than baseline.

A ratio lower than 1 means the experimental group outcome is lower than baseline.

For practical examples of how to analyze the journal articles outcomes, please refer to our journal club section. Articles with the RCT or therapy tags should allow trainees to practice understanding outcomes.