Article type
Year
Abstract
Background:Activities that try to fill up the “knowledge to action gap” can be considered as a Knowledge Translation(KT) intervention.Most of the KT interventions do not have enough compatibility, usually because of different levels of methodological quality over individual studies. It, most of the times, is responsible for observed heterogeneity in estimated effects among individual studies. Therefore, applying methods to reduce risk of bias(ROB) in individual KT studies can increase comparability and promotes validity of the conclusions of the related review studies.
Objectives:1-To describe how different methods (sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting) were applied to reduce ROB in KT interventional studies 2-To estimate the contribution of ROB to observed heterogeneity among individual study' estimated effects of KT interventions in health policy decision making.
Methods:Systematic review and meta-epidemiology design. Participants: policy makers. Intervention: structural, managerial or individual strategies. Outcomes: measures of research knowledge uptaking. Type of studies: randomized and non- randomized control trials. Search: MEDLINE, Cochrane central register, DARE, SCOPUS, Web of Science(search strategy is accessible in attached file1). Strategy for synthesis: narrative synthesis of the findings from the included studies which was structured around the type of the methodological technique and type of study intervention, outcome (according to EPOC) and audience was provided. Scores for each ROB(selection bias, performance bias, attrition bias, detection bias) and the overall ROB for every single included study were obtained. The relationship between the type of intervention/method/audience with ROB was estimated via meta-epidemiology analysis.The contribution of each method to the observed heterogeneity among individual studies were estimated through meta-regression.
Results:Out of 1633 retrieved independent studies, 31 were eligible and 17 were included in quantitative analysis. risk of bias was higher in group level studies, consensus process intervention and non-randomised trial. Studies which applied Sequence generation techniques in intervention allocation process showed lower total score(Mean(sd): 38(18) vs 76(31)). Meta-epidemiology analysis showed that SMD in studies with higher ROB was 0.17(CI:0.05- 0.29) greater than studies with lower ROB over all types of KT interventions(figure 1).Studies that showed stronger effect have higher risk of bias in meta-regression analysis(B=.013, p.007)(figure2).
Conclusion: Risk of bias can distort observed studies result in a way that they would show more exaggerated resulted values.This distortion is seeming higher in more complex interventions than simple interventions and when there are higher levels of subjectivity in studies measures.
Patient or healthcare consumer involvement:Policy makers have shared their opinion on the study' objectives and the results interpretation so they have been modified accordingly.
Objectives:1-To describe how different methods (sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting) were applied to reduce ROB in KT interventional studies 2-To estimate the contribution of ROB to observed heterogeneity among individual study' estimated effects of KT interventions in health policy decision making.
Methods:Systematic review and meta-epidemiology design. Participants: policy makers. Intervention: structural, managerial or individual strategies. Outcomes: measures of research knowledge uptaking. Type of studies: randomized and non- randomized control trials. Search: MEDLINE, Cochrane central register, DARE, SCOPUS, Web of Science(search strategy is accessible in attached file1). Strategy for synthesis: narrative synthesis of the findings from the included studies which was structured around the type of the methodological technique and type of study intervention, outcome (according to EPOC) and audience was provided. Scores for each ROB(selection bias, performance bias, attrition bias, detection bias) and the overall ROB for every single included study were obtained. The relationship between the type of intervention/method/audience with ROB was estimated via meta-epidemiology analysis.The contribution of each method to the observed heterogeneity among individual studies were estimated through meta-regression.
Results:Out of 1633 retrieved independent studies, 31 were eligible and 17 were included in quantitative analysis. risk of bias was higher in group level studies, consensus process intervention and non-randomised trial. Studies which applied Sequence generation techniques in intervention allocation process showed lower total score(Mean(sd): 38(18) vs 76(31)). Meta-epidemiology analysis showed that SMD in studies with higher ROB was 0.17(CI:0.05- 0.29) greater than studies with lower ROB over all types of KT interventions(figure 1).Studies that showed stronger effect have higher risk of bias in meta-regression analysis(B=.013, p.007)(figure2).
Conclusion: Risk of bias can distort observed studies result in a way that they would show more exaggerated resulted values.This distortion is seeming higher in more complex interventions than simple interventions and when there are higher levels of subjectivity in studies measures.
Patient or healthcare consumer involvement:Policy makers have shared their opinion on the study' objectives and the results interpretation so they have been modified accordingly.