Table 1.

ASTRO recommendation grading classification system

ASTRO’s recommendations are based on evaluation of multiple factors including the QoE, individual study quality, and panel consensus, all of which inform the strength of recommendation. QoE is based on the body of evidence available for a particular key question and includes consideration of number of studies, study design, adequacy of sample sizes, consistency of findings across studies, and generalizability of samples, settings, and treatments.
Strength of Recommendation	Definition		Overall QoE Grade	Recommendation Wording
Strong	•	• Benefits clearly outweigh risks and burden, or risks and burden clearly outweigh benefits.	Any (usually high, moderate, or expert opinion)	“Recommend/Should”
	•	• All or almost all informed people would make the recommended choice.
Conditional	•	• Benefits are finely balanced with risks and burden or appreciable uncertainty exists about the magnitude of benefits and risks.	Any (usually moderate, low, or expert opinion)	“Conditionally Recommend”
	•	• Most informed people would choose the recommended course of action, but a substantial number would not.
	•	• A shared decision-making approach regarding patient values and preferences is particularly important.
Overall QoE Grade	Type/Quality of Study		Evidence Interpretation
High	•	• 2 or more well-conducted and highly generalizable RCTs or meta-analyses of such trials.	The true effect is very likely to lie close to the estimate of the effect based on the body of evidence.
Moderate	•	• 1 well-conducted and highly generalizable RCT or a meta-analysis of such trialsOR	The true effect is likely to be close to the estimate of the effect based on the body of evidence, but it is possible that it is substantially different.
	•	• 2 or more RCTs with some weaknesses of procedure or generalizabilityOR
	•	•2 or more strong observational studies with consistent findings.
Low	•	• 1 RCT with some weaknesses of procedure or generalizabilityOR	The true effect may be substantially different from the estimate of the effect. There is a risk that future research may significantly alter the estimate of the effect size or the interpretation of the results.
	•	• 1 or more RCTs with serious deficiencies of procedure or generalizability or extremely small sample sizesOR
	•	• 2 or more observational studies with inconsistent findings, small sample sizes, or other problems that potentially confound interpretation of data.
Expert Opinion*	•	• Consensus of the panel based on clinical judgment and experience, due to absence of evidence or limitations in evidence.	Strong consensus (≥90%) of the panel guides the recommendation despite insufficient evidence to discern the true magnitude and direction of the net effect. Further research may better inform the topic.

Abbreviations: ASTRO = American Society for Radiation Oncology; QoE = quality of evidence; RCT = randomized controlled trial.

^*A lower quality of evidence, including expert opinion, does not imply that the recommendation is conditional. Many important clinical questions addressed in guidelines do not lend themselves to clinical trials but there still may be consensus that the benefits of a treatment or diagnostic test clearly outweigh its risks and burden.