Abstract: FR-PO1198
Reporting and Handling of Missing Outcome Data in Systemic Reviews of Kidney Transplant Studies
Session Information
- Transplantation: Clinical - Immunosuppression, Adherence, Outcomes
November 08, 2019 | Location: Exhibit Hall, Walter E. Washington Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Transplantation
- 1902 Transplantation: Clinical
Authors
- Budhiraja, Pooja, University of Kansas, Kansas city, Kansas, United States
- Kalot, Mohamad A., University of Kansas, Kansas city, Kansas, United States
- El alayli, Abdallah, Lebanese American University Medical Center, Beirut, Lebanon
- Dimassi, Ahmad Bilal, Lebanese American University Medical Center, Beirut, Lebanon
- Ilahe, Amna, University of Kansas, Kansas city, Kansas, United States
- Mustafa, Reem, University of Kansas, Kansas city, Kansas, United States
Background
Missing outcome data (MOD) can be absent not at random but due to side effects or in effectivity of interventions and can have implications on the validity, reproducibility and generalizability of the results. Hence, before concluding the effectiveness of the intervention in clinical setting it is important for the systematic reviews (SRs) to collect information about MOD and perform appropriate analysis to assess the robustness of the results.
Methods
We conducted a methodological survey of reporting and handling of MOD in SRs published in past 5 years. We included meta-analyses of randomized controlled trials performed in adult kidney transplant recipients that provided pooled estimate of an intervention on at least one dichotomous outcome. We used a standardized pilot tested forms with detailed instructions for each step of the review process. Title, abstract, full text screening and data abstraction all are done in duplicates. We studied how SRs collected, reported and handled MOD in the primary analysis.
Results
Seventy-one SRs (14 Cochrane and 57 Non-Cochrane reviews) met the inclusion criteria. Drugs were the most common intervention studied (84%) and average follow up was 12 months. Intention to Treat (ITT) or modified ITT was reported in 50% Cochrane and 9% non-Cochrane reviews. Ninety one percent of SRs did not collect information for number or reasons for participants with MOD. Only 5% handled and justified the analytical method(s) used to handle MOD. Furthermore, only 4% performed sensitivity analysis to account for MOD and considered the uncertainty associated with imputing outcomes in the analysis. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach for evaluation of the confidence in estimates of effect was used in 21% of the SRs.
Conclusion
Missing outcome data can introduce bias due to systematic differences between the observed and unobserved data, which can compromise the certainty in the evidence. SRs in kidney transplant recipients do not adequately report, handle or discuss the risk of bias associated with the MOD before concluding the results.