Developing a clinical practice guideline is a lengthy process, involving multiple interest-holders and complex methodology. Internationally accepted standards necessitate guidelines be based on rigorous systematic reviews of the evidence [1,2,3,4,5]. In 25 years, the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach has proven to be a robust and transparent system that rates the certainty (or quality) of evidence and establishes a framework to determine direction and strength of recommendations [6, 7].

Building on prior editorials on certainty and guideline adoption [8, 9], this report outlines how GRADE guides evidence to recommendations and adaptation, helping ophthalmologists understand their rationale and application in clinical practice.

Evidence-to-Decision (EtD) frameworks

GRADE offers guidance in moving from evidence to recommendations through the Evidence-to-Decision (EtD) Frameworks—comprised of all factors that may reasonably influence a recommendation—not only evidence of health benefits and harms, but also values, resources, equity, acceptability, and feasibility. For ophthalmologists interpreting/applying guidelines, the presence of EtD frameworks signals that the guideline panel likely engaged in a transparent deliberation of all considerations to formulate recommendations. Reviewing the framework enables clinicians to understand each recommendation’s rationale.

The EtD framework contains seven main factors: balance between desirable and undesirable effects, certainty of evidence, values and preferences, resource use, equity, acceptability, and feasibility [10]. (Table 1) For ophthalmologists, understanding how panels considered these domains offers insight into the reasoning behind recommendations.

Table 1 Evidence-to-Decision framework factors and strength of recommendation.
Full size table

When formulating recommendations using the EtD framework, panels begin by evaluating whether an intervention meaningfully improves outcomes that matter to patients—such as visual acuity, pain, or quality of life—while minimizing harms. Panels also consider the certainty of the evidence. This judgment incorporates factors such as risk of bias, consistency across studies, precision, and directness of the evidence. According to GRADE, high or moderate certainty evidence suggests that the estimated effect is close to the true effect.

Patient values and preferences—often neglected in guidelines—capture the degree to which patients, if fully informed, would judge that the benefits of a particular intervention outweigh its harms or vice versa. Most patients might prefer topical fluoroquinolones over no antibiotics, as they increase clinical cure rates (123 more per 1000) from bacterial conjunctivitis when the risk of adverse events is negligible (21 fewer per 1000) [11]. However, if the same intervention increases the risk of adverse events meaningfully (e.g., to 50 more per 1000), preferences may diverge. Significant uncertainty and/or variability in values and preferences necessitate shared decision-making [12,13,14].

Finally, EtD framework directs panels to consider contextual factors, such as resource use, equity, acceptability, and feasibility. While these factors should be considered for every guideline, their emphasis may vary [15]. In a patient-centred guideline, treatment costs might be critical in a private healthcare system but less important in a publicly funded health system with full drug coverage.

Direction and strength of recommendations

GRADE provides guideline developers a system for deciding on the direction and strength of recommendations. Guidelines provide recommendations in favor of an intervention when the desirable benefits—along with contextual factors—favor the intervention, based on patients’ values and preferences. Conversely, they recommend against an intervention when its potential harms outweigh benefits.

Historically, guidelines have used a combination of letters and numbers (i.e., I, IIa, IIb), but their variable use across guidelines is prone to misinterpretation. Guidelines that appropriately apply GRADE rate the strength of recommendations in two categories: strong or conditional/weak.

Guidelines developed using GRADE issue strong recommendations when there is a clear imbalance between benefits and harms—considering both health and contextual factors—based on patients’ values and preferences and supported by high or moderate certainty evidence. Conversely, guidelines offer conditional recommendations when the benefits outweigh harms or vice versa, for most patients, or when the evidence is of low or very low certainty. Ophthalmologists should be apprehensive of strong recommendations based on lower certainty evidence. In a Royal College of Ophthalmologists guideline on idiopathic full-thickness macular holes, the panel avoided strong recommendations for the treatment of holes that don’t close after surgery due to low certainty evidence [16]. On the other hand, due to high certainty evidence of large net benefit and predictable patient preference, the American Academy of Ophthalmology issued a strong recommendation in favor of anti-VEGF agents for patients with age-related macular disease [17].

From the perspective of ophthalmologists, clinicians can apply strong recommendations without hesitation; however, conditional recommendations signal the need for shared decision-making with patients. To optimize clarity, GRADE suggests standard language to enable immediate recognition of the strength of recommendations [18]. While strong recommendations typically begin with “We recommend (…)”, weak ones use “We suggest (…)”.

Examples below illustrate how recommendations are usually presented using GRADE framework:

“On stable therapy, ophthalmic monitoring no less frequently than every 3 months is strongly recommended over monitoring less frequently” [19].

“The PHC/Adult Hospital Level Committee suggests using methotrexate for the management of non-infectious posterior uveitis (…)” [20].

GRADE-Adolopment

Aware of the need to adapt recommendations to different contexts, GRADE also offers a framework called “Adolopment” [21, 22]. This approach allows guideline developers to adopt or adapt existing recommendations or formulate de novo recommendations (Fig. 1).

Fig. 1
figure 1

GRADE-Adolopment flowchart.

Full size image

Panelists identify relevant guidelines and assess whether the evidence supporting available recommendations remains current and applicable. Panels then evaluate how other considerations, as outlined in EtD frameworks, might warrant modification, based on the local context.

The importance of this contextualization becomes clear in examples such as the World Health Organization’s recommendation for empiric use of antibiotics in suspected bacterial keratitis [23]. In lower-resource settings where access to ophthalmologists may be limited, primary care physicians offering topical antibiotics is pragmatic. However, in settings with ready specialist access, adaptation may favor early referral, recognizing the need for specialist management to optimize outcomes. For ophthalmologists, Adolopment principles help interpret/apply international recommendations to local settings.

In another guideline, recommendations for Juvenile Arthritis-associated uveitis were contextualized to the Canadian context using GRADE-Adolopment—five adapted, two omitted and one de novo, reflecting differences in access, equity, and resource constraints [24].

Conclusion

GRADE system, supported by the EtD framework, offers a transparent approach for formulating recommendations, while GRADE-Adolopment extends this guiding adaptation to local contexts.