Appendix 3: Grading of Evidence

Levels of evidence for studies of intervention

Ia
From systematic review and meta-analysis of randomised controlled trials (RCTs)
Ib
Individual RCT(s) (with narrow confidence intervals)
IIa
Systematic review of at least one non-randomised controlled trial or well-designed cohort study
IIb
Individual cohort study or low quality RCTs
IIIa
Systematic review of at least one case-controlled study
IIIb
Individual case-control study
IV
Expert committee reports or opinions and/or clinical experience of authorities, case series (and poor-quality cohort and case-control studies)

Levels of evidence for validity of candidate risk factors

Ia
Systematic reviews or meta-analysis of level I studies with a high degree of homogeneity
Ib
Systematic reviews or meta-analysis with moderate or poor homogeneity
Ic
Level I studies (with appropriate populations and internal controls)
IIa
Systematic reviews or meta-analysis of level II studies
IIb
Level II studies (inappropriate population or lacking an internal control)
IIIa
Systematic reviews or meta-analysis of level III studies
IIIb
Case-control studies
IV
Evidence from expert committees without explicit critical scientific analysis or that based on physiology, basic research or first principles.

Of note, FRAX risk factors are all grade A or B according to evidence for reversibility of risk 63.

Grading of recommendations

Recommendations follow the Grading of Recommendations Assessment, Development, and Evaluation GRADE binary classification of recommendations as either strong or conditional (also known as discretionary or qualified recommendations) 311. Recommendations have been made after assessment of 312:

  1. The balance between desirable and undesirable effects -The larger the difference between the desirable and undesirable effects, the more likely a strong recommendation is warranted.
  2. The quality of evidence - The higher the quality of evidence, the more likely a strong recommendation is warranted.
  3. Values and preferences - The more variability/ uncertainty in values and preferences the more likely a conditional recommendation is warranted.
  4. Costs (resource allocation) - The higher the costs of an intervention (i.e., the more resources consumed) the more likely a conditional recommendation is warranted.

For example, a strong recommendation applies where the clinician considers that most people ought to receive the intervention, or where adherence to the recommendation could be used as a performance or quality indicator and that deviation from this recommendation would prompt documentation of a clinician’s rationale. NICE suggests using ‘offer’ (or similar action wording such as ‘measure’, ‘advise’, ‘commission’ or ‘refer’) when describing a strong recommendation313.

A conditional recommendation applies where the clinician examines the evidence and prepares to discuss this with the patient together with the patient’s values and preferences, or where documentation of the discussion of the pros and cons of an intervention is the indicator of quality, rather than the course of action itself. NICE suggests using wording such as ‘consider’ when describing conditional recommendations.

Where insufficient evidence is available or the evidence available is equivocal, recommendations are not made.