I

At least one properly done RCT

II-1

Well-designed controlled trial without randomization

II-2

Well-designed cohort or case-control analytic study, preferably from more than one source

II-3

Multiple time series evidence with/without intervention, dramatic results of uncontrolled experiment

III

Opinion of respected authorities, descriptive studies, case reports, and expert committees

Good

High grade evidence (I or II-1) directly linked to health outcome

Fair

High grade evidence (I or II-1) linked to intermediate outcome;
or
Moderate grade evidence (II-2 or II-3) directly linked to health outcome

Poor

Level III evidence or no linkage of evidence to health outcome

Substantial

More than a small relative impact on a frequent condition with a substantial burden of suffering;
or
A large impact on an infrequent condition with a significant impact on the individual patient level.

Moderate

A small relative impact on a frequent condition with a substantial burden of suffering;
or
A moderate impact on an infrequent condition with a significant impact on the individual patient level.

Small

A negligible relative impact on a frequent condition with a substantial burden of suffering;
or
A small impact on an infrequent condition with a significant impact on the individual patient level.

Zero or Negative

Negative impact on patients;
or
No relative impact on either a frequent condition with a substantial burden of suffering; or an infrequent condition with a significant impact on the individual patient level.

A strong recommendation that the clinicians provide the intervention to eligible patients.
Good evidence was found that the intervention improves important health outcomes; the conclusion is made that benefits substantially outweigh harm.

A recommendation that clinicians provide this intervention to eligible patients.
At least fair evidence was found that the intervention improves health outcomes; the conclusion is made that benefits outweigh harm.

No recommendation for or against the routine provision of the intervention is made.
At least fair evidence was found that the intervention can improve health outcomes, but benefits and harms are too closely balanced to justify a general recommendation.

Recommendation is made against routinely providing the intervention to asymptomatic patients.
At least fair evidence was found that the intervention is ineffective or the conclusion is made that harms outweigh benefits.

Evidence is insufficient to recommend for or against routinely providing the intervention.
Evidence shows that the effectiveness of the intervention lacking, is of poor quality, or is conflicting;  the conclusion is that the balance of benefits and harms cannot be determined.