Article type
Year
Abstract
Background: Estimating absolute treatment effects requires establishment of baseline risk (control event rate) and then application of relative risks of interventions to that baseline risk. Systematic reviews of baseline risk are not common, and application of GRADE Working Group guidance to such reviews even less common. Methodologic challenges, and the necessity for innovative solutions, arise in such situations, one of which is establishing baseline risk of venous thromboembolism (VTE) and bleeding in patients undergoing urologic surgery.
Objectives: To address challenges encountered in a series of systematic reviews addressing the trade-off between benefits (VTE prevention) and risks (bleeding) of thromboprophylaxis in patients undergoing urologic surgical procedures.
Methods: Review of relevant literature and systematic review team brainstorming, development of approaches, and iterative testing and refinement.
Results: Challenges encountered and solutions adopted included the following:
1. identifying risk of bias issues most relevant to this setting;
2. variable duration of follow-up: we identified natural history studies that informed timing of VTE and bleeding and modeled the frequency of events accordingly;
3. choice of best estimate: when there were sufficient studies we chose the median of the higher quality studies, when few studies existed we chose the median of all studies;
4. variable use of prophylaxis: we modeled event rates using results of systematic reviews of RCT of the impact of prophylaxis on bleeding and thrombosis;
5. estimating deaths: we applied mortality estimates for VTE and bleeding from the studies with sufficient numbers of patients to make the estimates to those that did not;
6. risk stratification: we created a simple novel instrument for risk on the basis of available prognostic studies;
7. certainty in estimates: we adapted criteria for each GRADE domain to this context including generating 'certainty intervals' that incorporated limitations beyond chance in generating quantitative estimates of uncertainty.
Conclusions: Authors conducting reviews of baseline risk may benefit from our experience.
Objectives: To address challenges encountered in a series of systematic reviews addressing the trade-off between benefits (VTE prevention) and risks (bleeding) of thromboprophylaxis in patients undergoing urologic surgical procedures.
Methods: Review of relevant literature and systematic review team brainstorming, development of approaches, and iterative testing and refinement.
Results: Challenges encountered and solutions adopted included the following:
1. identifying risk of bias issues most relevant to this setting;
2. variable duration of follow-up: we identified natural history studies that informed timing of VTE and bleeding and modeled the frequency of events accordingly;
3. choice of best estimate: when there were sufficient studies we chose the median of the higher quality studies, when few studies existed we chose the median of all studies;
4. variable use of prophylaxis: we modeled event rates using results of systematic reviews of RCT of the impact of prophylaxis on bleeding and thrombosis;
5. estimating deaths: we applied mortality estimates for VTE and bleeding from the studies with sufficient numbers of patients to make the estimates to those that did not;
6. risk stratification: we created a simple novel instrument for risk on the basis of available prognostic studies;
7. certainty in estimates: we adapted criteria for each GRADE domain to this context including generating 'certainty intervals' that incorporated limitations beyond chance in generating quantitative estimates of uncertainty.
Conclusions: Authors conducting reviews of baseline risk may benefit from our experience.