Martin Saborido C1, Jimenez D2, Muriel A3, Zamora J3, Morillo R2, Barrios DD2, .... This communication presents all relevant information following The PRISMA ...
EFFICACY AND SAFETY OUTCOMES OF RECANALIZATION PROCEDURES IN PATIENTS WITH ACUTE SYMPTOMATIC PULMONARY EMBOLISM: SYSTEMATIC REVIEW AND NETWORK META-ANALYSIS Martin Saborido C1, Jimenez D2, Muriel A3, Zamora J3, Morillo R2, Barrios DD2, Klok E4, Huisman M4, Yusen R5, Tapson V6 1Universidad
Francisco de Vitoria, Pozuelo de Alarcón, Spain, 2Ramon y Cajal Hospital, IRYCIS and Alcala University, Madrid, Spain, 3Hospital Ramón y Cajal, Madrid, Spain, 4Leiden University, Leiden, The Netherlands, 5Washington University Physicians, St. Louis, MO, USA, 6Cedars Sinai, Los Angeles, CA, USA
This communication presents all relevant information following The PRISMA Extension Statement for Reporting of Systematic Reviews Incorporating Network Metaanalyses of Health Care Interventions to facilitate reading, understanding and critical assessment. The aim of this study is to review the efficacy and safety of the recanalization procedures for the treatment of acute symptomatic pulmonary embolism (PE). Protocol: Protocol registered in PROSPERO: CRD42015024670. Eligibility: Randomized clinical trials that compared the effect of a recanalization procedure vs. each other or anticoagulant therapy in patients diagnosed with acute symptomatic PE were included. The primary outcomes were all-cause mortality and major bleeding. Searches: Searches were performed in PubMed, the Cochrane Library, EMBASE, EBSCO, Web of Science, and CINAHL from inception through July 31, 2015, without language restrictions. Study selection: Two reviewers independently screened by abstract and full text and extracted data. Disagreements were resolved by consensus. Data extraction: Two reviewers independently extracted data onto a computer spreadsheet, with discrepancies resolved by consensus. Extracted data included first author, year of publication, type of intervention and control group, number of patients, patient characteristics, and duration of follow-up. The primary outcomes were all-cause mortality and major bleeding, as defined by the study protocol. Secondary outcomes were risk of intracranial hemorrhage (ICH) and recurrent embolism. Geometry of the network: Nodes describing single interventions. Risk of bias within studies: Risk of bias have been assessed at study level and synthesized graphically. Summary measures: primary and secondary outcomes have been summarized using OR including confidence intervals (CI). Treatment ranking have been elaborated using surface under the cumulative ranking curve (SUCRA) values. Bivariate analysis has been graphically included for primary outcomes. Assessment of consistency: Consistency has been evaluated using a loop specific approach and by fitting an inconsistency model. Risk of bias across studies: Publication bias has been assessed graphically by means of funnel plots. Additional analysis: We performed some sensitivity analyses to assess the robustness of the findings. These were based on (1) restricting only to studies in patients with hemodynamically stable PE; (2) restricting only to trials where the mean age of participants in the thrombolytic group was > 65 years; and (3) alternative statistical model (frequentist approach using a randomeffects inconsistency model). A bivariate analysis using the primary outcomes has been performed as well.
Study selection
Synthesis of results Pharmacological
Consistency
intervention
Hemodynamically stable
> 65 years
model
Inconsistency model
ALL-CAUSE MORTALITY Compared with anticoagulation Full-dose
0.60 (0.36-1.01)
0.59 (0.29-1.22)
0.75 (0.31-1.81)
0.61 (0.36-1.04)
0.47 (0.14-1.59)
1.13 (0.04-31.83)
0.36 (0.01-20.70)
0.58 (0.04-8.70)
0.31 (0.01-7.96)
0.31 (0.01-7.96)
-
0.55 (0.13-2.31)
thrombolysis Low-dose thrombolysis Catheter-directed thrombolysis Compared with full-dose thrombolysis Low-dose
1.28 (0.40-4.12)
1.90 (0.07-49.60)
0.48 (0.01-25.18)
0.89 (0.23-3.41)
1.93 (0.07-51.33)
0.53 (0.02-14.54)
-
0.51 (0.02-13.56)
Blinding of outcome assessment (detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias)
Compared with low-dose thrombolysis Catheter-directed
1.50 (0.05-47.94)
0.28 (0-29.11)
-
Other bias
0.67 (0.02-21.26)
0%
thrombolysis Low risk of bias
MAJOR BLEEDING
Unclear risk of bias
25%
50%
75%
100%
High risk of bias
Compared with anticoagulation Full-dose
2.00 (1.06-3.78)*
2.29 (0.78-6.71)
3.39 (1.01-11.42)* 1.99 (1.04-3.79)*
0.90 (0.25-3.21)
0.43 (0.01-18.62)
4.93 (0.30-80.95)
1.11 (0.02-78.75)
0.97 (0.02-56.03)
0.97 (0.01-66.76)
-
0.89 (0.23-3.41)
Risks of bias across studies
thrombolysis Low-dose
Network for all-case mortality and bleeding
Allocation concealment (selection bias) Blinding of participants and personnel (performance bias)
thrombolysis
The table to the right synthesizes the results and includes the consistency model and inconsistency model along with two sensitivity analyses. Statistical differences have been found only in major bleedings when compares fulldose thrombolysis and anticoagulation, in both the consistency and inconsistency model and in one sensitivity analysis (>65 years ).
Risk of bias within studies Random sequence generation (selection bias)
thrombolysis Catheter-directed
CONCLUSION and LIMITATIONS Recanalization procedures don’t seem to offer a clear advantage compared with anticoagulation for PE. Lowdose thrombolysis was associated with the lowest probability of dying and bleeding. Since some evidence suggests that quality of thrombolytic clinical trials has substantially changed in the past 30 years, we believe that inconsistency should be examined thoroughly using the different methods used in the older studies compared with those done more recently.
thrombolysis Catheter-directed thrombolysis Compared with full-dose thrombolysis Low-dose
2.22 (0.71-6.89)
0.19 (0.01-6.94)
1.45 (0.12-18.09)
0.45 (0.14-1.46)
2.07 (0.03-126.08)
0.42 (0.01-33.32)
-
0.49 (0.01-29.81)
-
1.09 (0.02-78.62)
thrombolysis Catheter-directed thrombolysis Compared with low-dose thrombolysis Catheter-directed
0.93 (0.01-65.65)
2.23 (0.01-640.48)
thrombolysis
Additional analysis: Bivariate analysis
The bivariate analysis completes the overall picture of “best profile” for the interventions. This is based on the results from the estimated ranking where catheter-directed thrombolysis was associated with the lowest probability of dying (SUCRA, 0.67), followed by low-dose thrombolysis (SUCRA, 0.66), and full-dose thrombolysis (SUCRA, 0.55). Similarly, low-dose thrombolysis was associated with the lowest probability of major bleeding (SUCRA. 0.61), followed by catheter-directed thrombolysis (SUCRA, 0.54), and fulldose thrombolysis (SUCRA, 0.17).
Loop inconsistency Loop IF
se IF
z-value p value CI 95%
A-B-C 0.746 1.373 0.544
0.587
Loop Heterogeneity tau
(0.00-3.44) 0.000
A: Full-dose thrombolysis B: Anticoagulation C: Low-dose thrombolysis D: Catheter-directed thrombolysis
There were significant differences between direct and indirect estimates in the only closed loop that allowed assessment of network consistency (anticoagulation-full-dose thrombolysis-low-dose thrombolysis).
