Recommendation #3: Be strategic about outcomes
Investigators conducting deprescribing trials often wish to determine the effect of their interventions on “big-ticket” outcomes such as mortality, hospitalization, and quality of life. We can do ourselves a disservice by reaching for these endpoints, as we will often fall short. Given how many factors outside of medication use impact these outcomes, deprescribing interventions may need to have unrealistically potent effects to detect a difference given the limited sample sizes we typically employ. Consider cardiology trials, which often enroll tens of thousands of patients over multiple years to try and achieve such outcomes. And, contrast this with drugs like ezetimibe, which was approved by the US Food and Drug Administration on the basis of beneficial effects on lipid levels without any data on cardiovascular morbidity or mortality.8 If this is the evidence required to generate widespread use of a drug with well over $1 billion in sales, we may be setting too ambitious a standard for ourselves to attain.
An interesting contrast is set by the OPTIMIZE trial, which used a non-inferiority design and was published in JAMA with substantial attention. This study showed that older adults with well-controlled blood pressure (mean systolic, 130 mm Hg) who stopped one antihypertensive medication had similar rates of remaining at a systolic blood pressure <150 mm Hg than people receiving usual care.9 This flips the perceived value of deprescribing on its head – rather than having to show that deprescribing improves clinical outcomes, in many settings it may be enough to say that people can fare just as well stopping their medications than continuing them. To be clear, we should not choose outcomes just because they are easy. However, we should pursue opportunities where outcomes are both clinically meaningful and have realistic potential to demonstrate benefit – or, where appropriate, non-inferiority - in response to interventions.