For recommendations to truly be called evidence-based, a system must go far beyond citing evidence to support the recommendation.
The evidence used to inform the recommendation must be systematically identified and evaluated to use the current comprehensive evidence with a critical understanding of objective results and reliability of conclusions. However, recommendations require applying value judgments to factual conclusions so “evidence-based recommendations” require consistent and systematic methods for determining recommendations based on the best available evidence to inform weighing desirable and undesirable consequences, and such judgments being applied by clinical experts without conflicts of interest. Synthesized recommendations provide a global perspective where recommendations have been presented by multiple groups.
Synthesized Recommendation Grading System for DynaMed Plus:
- DynaMed Plus systematically monitors clinical evidence to continuously provide a synthesis of the most valid relevant evidence to support clinical decision-making (See DynaMed Plus Evidence-based Methodology).
- Guideline recommendations summarized in the body of a DynaMed Plus topic are provided with the recommendation grading system used in the original guideline(s), and allow DynaMed Plus users to quickly see where guidelines agree and where guidelines differ from each other and from the current evidence.
- In DynaMed Plus, we synthesize the current evidence, current guidelines from leading authorities, and clinical expertise to provide recommendations to support clinical decision-making in the Overview & Recommendations section.
- We use the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system to classify synthesized recommendations as Strong or Weak.
- Strong recommendations are used when, based on the available evidence, clinicians (without conflicts of interest) consistently have a high degree of confidence that the desirable consequences (health benefits, decreased costs and burdens) do, or do not, outweigh the undesirable consequences (harms, costs, burdens).
- Weak recommendations are used when, based on the available evidence, clinicians believe that desirable and undesirable consequences are finely balanced, or appreciable uncertainty exists about the magnitude of expected consequences (benefits and harms). Weak recommendations are used when clinicians disagree in judgments of relative benefit and harm, or have limited confidence in their judgments. Weak recommendations are also used when the range of patient values and preferences suggests that informed patients are likely to make different choices.
- DynaMed Plus synthesized recommendations (in the Overview & Recommendations section) are determined with a systematic methodology:
- Recommendations are initially drafted by clinical editors (including ≥ 1 with methodological expertise and ≥ 1 with content domain expertise) aware of the best current evidence for benefits and harms, and the recommendations from guidelines.
- Recommendations are phrased to match the strength of recommendation. Strong recommendations use “should do” phrasing, or phrasing implying an expectation to perform the recommended action for most patients. Weak recommendations use “consider” or “suggested” phrasing.
- Recommendations are explicitly labeled as Strong recommendations or Weak recommendations when a qualified group has explicitly deliberated on making such a recommendation. Group deliberation may occur during guideline development. When group deliberation occurs through DynaMed-initiated groups:
- Clinical questions will be formulated using the PICO (Population, Intervention, Comparison, Outcome) framework for all outcomes of interest specific to the recommendation to be developed.
- Systematic searches will be conducted for any clinical questions where systematic searches were not already completed through DynaMed content development.
- Evidence will be summarized for recommendation panel review including for each outcome the relative importance of the outcome, the estimated effects comparing intervention and comparison, the sample size, and overall quality rating for the body of evidence.
- Recommendation panel members will be selected to include at least three members that together have sufficient clinical expertise for the subject(s) pertinent to the recommendation, methodological expertise for the evidence being considered, and experience with guideline development.
- All recommendation panel members must disclose any potential conflicts of interest (professional, intellectual, and financial), and will not be included for the specific panel if a significant conflict exists for the recommendation in question.
- Panel members will make Strong recommendations if and only if there is consistent agreement in a high confidence in the likelihood that desirable consequences outweigh undesirable consequences across the majority of expected patient values and preferences. Panel members with make Weak recommendations if there is limited confidence (or inconsistent assessment or dissenting opinions) that desirable consequences outweigh undesirable consequences across the majority of expected patient values and preferences. No recommendation is made if there is insufficient confidence to make a recommendation.
- All steps in this process (including evidence summaries which were shared with the panel, and identification of panel members) will be transparent and accessible in support of the recommendation.
- Recommendations are verified by ≥ 1 editor with methodological expertise, not involved in recommendation drafting or development, with explicit confirmation that Strong recommendations are adequately supported.
- Recommendations are published only after consensus is established with agreement in phrasing and strength of recommendation by all editors.
- If consensus cannot be reached then the recommendation can be published with a notation of “dissenting commentary” and the dissenting commentary is included in the topic details.
- If recommendations are questioned during peer review or post-publication by a qualified individual, or re-evaluation is warranted based on new information detected through systematic literature surveillance, the recommendation is subject to additional internal review.