Article type
Abstract
Background: Evidence-based practice states that effect estimates are closer to the truth when based on higher certainty evidence (CoE). Therefore, even if there is new evidence, the best estimates of treatment effects are less likely to change if they are based on high or moderate CoE, than if they are based on lower CoE.
Objectives: To describe how the CoE and the direction of effect estimates evolve as evidence accumulates in a living systematic review and network meta-analysis (LSRNMA). Secondarily, we will explore factors associated with changes in CoE or effect direction over time.
Methods: We will use direct and network estimates and GRADE assessments from each of the four published iterations of our LSRNMA of COVID-19 treatments. We will include data from comparisons of drug treatments to standard care/placebo for which there are results in at least two iterations of our LSRNMA. We will analyze changes for the outcomes of mortality, mechanical ventilation, adverse events, admission to hospital, duration of mechanical ventilation, hospital and ICU length of stay. We will conduct a multilevel analysis to determine whether very low/low CoE is more likely to change compared to moderate/high CoE over time, and we will explore factors that could trigger these changes. We will quantify change with respect to CoE and the direction of the effect.
Results: This work is in progress. We will illustrate the evolution of CoE over time by comparing it at two points and indicating how frequently it changes. We will dichotomize the CoE as very low/low or moderate/high CoE at time one, and we will present how frequently the direction of the effect remains or changes between beneficial, no effect, or harmful in those categories. Additionally, we will present an adjusted likelihood of changes over time between observations with very low/low and moderate/high CoE
Conclusions: Our conclusions will shed light on the dynamics of evidence accumulation over time in our LSRNMAs. Understanding the changes in CoE and the direction of the effect over time, as well as identifying factors contributing to these changes, may inform decisions regarding updates to systematic reviews.
Objectives: To describe how the CoE and the direction of effect estimates evolve as evidence accumulates in a living systematic review and network meta-analysis (LSRNMA). Secondarily, we will explore factors associated with changes in CoE or effect direction over time.
Methods: We will use direct and network estimates and GRADE assessments from each of the four published iterations of our LSRNMA of COVID-19 treatments. We will include data from comparisons of drug treatments to standard care/placebo for which there are results in at least two iterations of our LSRNMA. We will analyze changes for the outcomes of mortality, mechanical ventilation, adverse events, admission to hospital, duration of mechanical ventilation, hospital and ICU length of stay. We will conduct a multilevel analysis to determine whether very low/low CoE is more likely to change compared to moderate/high CoE over time, and we will explore factors that could trigger these changes. We will quantify change with respect to CoE and the direction of the effect.
Results: This work is in progress. We will illustrate the evolution of CoE over time by comparing it at two points and indicating how frequently it changes. We will dichotomize the CoE as very low/low or moderate/high CoE at time one, and we will present how frequently the direction of the effect remains or changes between beneficial, no effect, or harmful in those categories. Additionally, we will present an adjusted likelihood of changes over time between observations with very low/low and moderate/high CoE
Conclusions: Our conclusions will shed light on the dynamics of evidence accumulation over time in our LSRNMAs. Understanding the changes in CoE and the direction of the effect over time, as well as identifying factors contributing to these changes, may inform decisions regarding updates to systematic reviews.