A large study examined the role of a new medication X in patients with known atherosclerotic vascular disease. Study subjects were randomized into intervention (medication X) versus placebo groups and followed for a mean of 3.9 years. The following results were reported: Which of the following is the most accurate statement about the study results?
A)Administration of medication X is associated with hemorrhagic cerebrovascular accident
B)Medication X can potentially prevent revascularization procedures
C)Number needed to treat to prevent 1 myocardial infarction is 25
D)The effect of the intervention is exaggerated due to intention-to-treat analysis
E)The intervention group had statistically significantly fewer major vascular events

Question

A large study examined the role of a new medication X in patients with known atherosclerotic vascular disease.  Study subjects were randomized into intervention (medication X) versus placebo groups and followed for a mean of 3.9 years.  The following results were reported:  Which of the following is the most accurate statement about the study results?
A)Administration of medication X is associated with hemorrhagic cerebrovascular accident 
B)Medication X can potentially prevent revascularization procedures 
C)Number needed to treat to prevent 1 myocardial infarction is 25 
D)The effect of the intervention is exaggerated due to intention-to-treat analysis 
E)The intervention group had statistically significantly fewer major vascular events

Quizplus · Accepted Answer

Explanation: For each major cardiovascular event, the rate ratio is represented by a square and its 95% confidence interval (CI) is represented by a horizontal line. In addition, the CIs and P-values for the overall categories (eg, "any" myocardial infarction, "any" cerebrovascular accident) but not the subcategories (eg, "nonhemorrhagic" and "hemorrhagic" cerebrovascular accident) are also written out. A single participant may experience >1 cardiovascular event, so there is some nonadditivity in the total events reported. Fewer participants in the medication X group underwent revascularization procedures (5.9% compared to 6.3% in the placebo group). The rate ratio for this was 0.89 and the corresponding CI (0.82-0.94) was the only one that did not cross the solid vertical line (gray color) that represents the null value (rate ratio of 1.0). This means that the effect is statistically significant (which is also suggested by the P-value of 0.03 <α = 0.05) and that the rate ratio is in favor of medication X (arrow pointing to the left). Therefore, medication X is potentially beneficial for the prevention of revascularization procedures. (Choice A) Although the risk of hemorrhagic cerebrovascular accident was higher in the medication X group (0.8%) than in the placebo group (0.6%), the CI crosses the solid vertical line (gray color), meaning that the difference between the groups is not statistically significant. Therefore, there is no statistically significant association between medication X use and hemorrhagic cerebrovascular accident. (Choice C) The number needed to treat (NNT) is calculated as 1/ARR, where ARR is the absolute risk reduction. For myocardial infarction, the ARR = 5.0% − 4.9% = 0.1% = 0.001, so NNT = 1/0.001 = 1000. (Choice D) Intention-to-treat analysis analyzes each participant based on randomization group (even if the participant stops the treatment or takes a different treatment). This tends to provide a conservative (rather than exaggerated) estimate of treatment effect. (Choice E) In the last row of the table, 14.3% of participants in the medication X group and 14.8% in the placebo group had a major vascular event; the rate ratio was 0.92. However, the corresponding CI (0.85−1.03) crosses the null value of 1.0 and the corresponding P-value is >0.05; this means that the difference between the groups is not statistically significant. Educational objective: A confidence interval that crosses the null value or a P-value >α error cutoff (typically 0.05) denotes a result that is not statistically significant. Intention-to-treat analysis tends to provide a conservative estimate of the effect of an intervention. References: Intention-to-treat concept: A review. (http://www.ncbi.nlm.nih.gov/pubmed/21897887)

A Large Study Examined the Role of a New Medication