Three parts of a defensible recommendation
A recommendation that can be examined and challenged states three things explicitly.
- Direction: recommend for, recommend against, or recommend either option.
- Strength: strong (most informed patients would choose it) or weak/conditional (the right choice depends on the patient's values and circumstances).
- Reasoning: the certainty of evidence, the balance of benefits and harms, the values and preferences at stake, and any resource considerations.
A recommendation with its reasoning
Using the low-certainty body of evidence from the previous page: 'We suggest offering the drug to patients with established cardiovascular disease (weak recommendation, low-certainty evidence). The pooled estimate suggests a reduction in the composite cardiovascular outcome, but the evidence is limited by unclear allocation concealment in two trials, unexplained heterogeneity, and possible publication bias. The absolute benefit depends on the patient's baseline risk and is likely small for lower-risk patients. Because the benefit is uncertain and the drug carries a small risk of adverse effects, the decision should be made with the patient, taking into account their baseline risk and their tolerance for uncertainty.' Notice that the direction is for, the strength is weak, and the reasoning names the certainty rating and the specific problems that produced it.
Strength is not the same as certainty
A strong recommendation can rest on low-certainty evidence when the consequences of being wrong are asymmetric — for example, a treatment for a condition that is otherwise uniformly fatal. Conversely, a weak recommendation can rest on high-certainty evidence when benefits and harms are closely balanced or when patient values differ substantially. Do not read a strong recommendation as a claim that the evidence is strong.