Measures of Effect: How Do We Interpret It from Evidence?

Relative effect measures express the expected event in one group compared to another group. The relative risk (RR) is the ratio between the risk of an event in two groups, while the odds ratio (OR) is the ratio of the ODDS of an event. For both measures, a value of 1 indicates that the estimate of the effect is the same in the two interventions.¹ For ease of understanding, we will take as an example the data from the RALES study (Randomized Aldactone Evaluation Study) shown in Table 1. This was a randomized clinical trial that randomized patients with a heart failure to spironolactone (intervention) versus a placebo (without intervention). The primary endpoint of this study was the mortality.² If we want to know the relative risk, we must first calculate the risk in the exposed or intervention group, that is, the ratio between the number of patients who died in the spironolactone group (226) over the total number of patients treated with this drug: 822 patients (dead + alive). The result of the ratio is 0.27, which is equivalent to saying that the risk of death in patients randomized to spironolactone was 27%. Then, we repeated the same procedure to calculate the risk of death in the patients in the placebo group: 314/841 = 0.37. The risk of death in the placebo group was 37%. Finally, to calculate the RR, the ratio of the risk of death in the spironolactone group over the risk of death in the placebo group was 0.27/0.37 = 0.72. A relative risk of 0.72 means that, in HF, spironolactone (compared to placebo) reduces mortality by 28%. If we are now interested in calculating the OR, it is first necessary to know the ODDS in the spironolactone group and then in the placebo group. For the first step, the ratio would be 226/596 = 0.37 and for the second, 314/527 = 0.59. The OR is the division of both, 0.37/0.59 = 0.62. An OR of 0.62 is interpreted as that for every 0.62 deaths that occur with spironolactone, 1 death occurs with placebo or every 62 deaths that occur with spironolactone, 100 occur with placebo.

Table 1. 2 x 2 table showing the mortality in the patients with spironolactone vs. with a placebo

Table 1
2 x 2 table showing the mortality in the patients with spironolactone vs with a placebo

Table 2 shows data from the PARTNER study with a 2-year follow-up, which randomized patients with severe aortic stenosis and high surgical risk to TAVR (Transcatheter aortic-valve replacement) or standard management. The primary event was defined as death and rehospitalizations.³ Following the same steps as in the previous example, if we calculate the RR and the OR, the results are 0.60 and 0.31. As can be seen, the effect measures are very similar in the spironolactone example and very different in the case of the percutaneous aortic valve intervention study. It is important to note that, when the event in question is frequent, the OR overestimates the effect and it is convenient to express the data as a RR.⁴

Table 2
2 x 2 table: combined event of death or rehospitalizations in patients with TAVR vs. with a placebo