Article type
Abstract
Background: The GRADE working group defines patient values and preferences as how patients value the relative importance of the main health outcomes. Although the GRADE working group has developed approaches to rating certainty of evidence treatment, diagnosis, resource and prognosis questions, guidance for assessing evidence regarding values and preferences thus far has been lacking.
Objectives: To provide guidance on how users can assess the certainty of evidence regarding importance of outcomes.
Methods: We applied the GRADE domains to rate several systematic reviews addressing importance of outcomes, conducted consensus meetings, and consulted stakeholders in the GRADE working group for feedback.
Results: A body of evidence addressing the importance of outcomes starts at 'high certainty'. Risk of bias, indirectness, inconsistency, imprecision and publication bias can lead to rating down this evidence. For risk-of-bias assessment, we propose subdomains of the selection of the study population, missing data, type of measurement tool, and confounding. We have also developed corresponding items for each subdomain. The population, intervention, comparison and outcome (PICO) elements of the rated evidence and methodological aspects determine the degree of indirectness. Inconsistency about typical values is generally due to PICO and methodological elements that should be explored and, if possible, like for other types of evidence, explained. The width of the confidence interval and sample size should inform judgments about imprecision. We also provide suggestions on how to detect publication bias based on empirical information. We also suggest within-study variability as a separate issue to the certainty of the evidence about typical values.
Conclusions: We have developed GRADE guidance for rating the certainty of evidence on how patients value health outcomes. This guidance will be helpful to systematic reviewers and decision makers, including guideline developers.
Objectives: To provide guidance on how users can assess the certainty of evidence regarding importance of outcomes.
Methods: We applied the GRADE domains to rate several systematic reviews addressing importance of outcomes, conducted consensus meetings, and consulted stakeholders in the GRADE working group for feedback.
Results: A body of evidence addressing the importance of outcomes starts at 'high certainty'. Risk of bias, indirectness, inconsistency, imprecision and publication bias can lead to rating down this evidence. For risk-of-bias assessment, we propose subdomains of the selection of the study population, missing data, type of measurement tool, and confounding. We have also developed corresponding items for each subdomain. The population, intervention, comparison and outcome (PICO) elements of the rated evidence and methodological aspects determine the degree of indirectness. Inconsistency about typical values is generally due to PICO and methodological elements that should be explored and, if possible, like for other types of evidence, explained. The width of the confidence interval and sample size should inform judgments about imprecision. We also provide suggestions on how to detect publication bias based on empirical information. We also suggest within-study variability as a separate issue to the certainty of the evidence about typical values.
Conclusions: We have developed GRADE guidance for rating the certainty of evidence on how patients value health outcomes. This guidance will be helpful to systematic reviewers and decision makers, including guideline developers.